Physics Suite PDF

Preface P-1 8/7/02:7:11 PM
of Maryland in 1970. I have been on the Maryland faculty ever since,

Preface teaching and doing research. From the first I had a strong interest in
teaching. From the first, I cheerfully ignored my colleagues advice to put a
A preface is the place where you expect to have some questions answered:
minimal effort into my teaching duties, since they would play little role in
What is this book about? Who is the intended audience? What value might
helping me get tenure.
I expect to get out of it?
In the 1980s I worked on trying to get the newly-invented personal
Although this book is associated with The Physics Suite and is distributed
computer into my classes. But as the decade went on, I became
as an instructor’s guide to using The Physics Suite, it really tries to be more
increasingly aware of two important facts: First, that my students were
than just a how to book for a particular set of materials. This book is about
learning to be a more effective teacher. having trouble learning physics  both with and without the computer 
and that their problems were more difficult to resolve than I had expected.
The intended audience is any physics teacher who is interested in learning Second, that there was a community that knew this and was studying it as a
about recent developments in physics education. In contrast to some other research effort. In 1991, I stopped doing nuclear physics and switched my
teachers’ guides, it is not a review of specific topics in physics with hints research activity to PER.
for how to teach them and lists of what common student difficulties are.
Rather, it is a handbook with a variety of tools for improving both teaching This history determines the structure of the book. Some of what I have
learned in 30 years of teaching has been from the research literature in PER
and learning of physics  from new kinds of homework and exam
and from my own work as an active physics education researcher. But a lot
problems, to surveys for figuring out what has happened in your class, to
has been from listening to and working with the students in my physics
tools for taking and analyzing data using computers and video.
classes over all those years. As a result, some of what helps my teaching is
Over the past two decades, a community has grown up in the scholarship of well documented through published research, but some is not.
teaching and learning, bridging physics and education. The people in this
Therefore, I have chosen to present this book as neither a research
community, which I refer to as Physics Education Research (PER), have
monograph nor as a standard didactic “how-to” teaching guide. Rather, I
tried to understand why so many students have so much difficulty
have decided to make it a “teacher-to-teacher” discussion in which I present
understanding physics, and they have tried to develop learning
what I have learned in three ways: as research results with data and citations
environments to help those students. The result has been a large body of
where they are available, as illustrative anecdotal examples from my own
knowledge and a growing repertoire of curricula that are demonstrably more
experience, and as general principles, guidelines, and heuristics that I have
effective than our traditional approaches.
found helpful.
The Physics Suite integrates materials from an active group of PER
The anecdotal examples do not always follow the events exactly.
developers. All the parts of The Physics Suite are based on education
Sometimes multiple stories have been combined and details omitted for
research and share a specific underlying philosophy. This book, while
clarity. Although these stories are based on real events, they should be
providing an introduction to the materials of the Physics Suite, more
interpreted as fables with morals rather than as records of real events.
importantly, provides an introduction to the educational philosophy and
knowledge base that form the foundation underlying the Suite. Because this I have tried to limit my general principles to ones that can be supported in
educational philosophy and knowledge base rest on well-documented three ways: by observations of real student behavior in real classrooms
scholarship, the foundation is broadly applicable. This foundation and this (usually by educational researchers), by controlled experimental studies on
book can help you to teach better even if you do not adopt a single item how people think (usually by cognitive scientists), and by physiological
from The Physics Suite. plausibility (consistency with what is known in neuroscience). Heuristics
(such as “Redish’s Teaching Commandments”) are less well-documented
Because the character of the book arises in a very personal way from my
and based on my own experience and what I have learned from other
own experiences, let me introduce myself briefly. I was trained as a
physics teachers.
theoretical nuclear physicist and began my teaching career at the University
Teaching Physics with the Physics Suite E. F. Redish

Preface P-2 8/7/02:7:11 PM
Throughout my career as a research physicist, my work had a strongly start in learning how to do PER and a view of what a PER group inside a
theoretical bias. I have always been interested in trying to understand how physics department looked like.
to think about and organize our knowledge of the real world. For trying to
Others whose work had a primary influence on my thinking include the late
understand the system of students trying to learn physics, the appropriate
Arnold Arons, John Clement, Fred Goldberg, David Hammer, Pat Heller,
theory to help us parse what we see into something sensible is cognitive
David Hestenes, Jose Mestre, and Fred Reif. I also want to thank the
science. As a result, this book has a strong cognitive flavor. Though I do
students, postdocs, and visitors I have worked with in PER at Maryland. I
not want to write a textbook in cognitive science, I have tried to extract and
have learned much from them and discussions with them have helped me
make plausible for physicists what is relevant and known in this area. For
clarify and refine my thinking on many occasions: in alphabetical order,
those who want more documentation or to understand better the strengths
Jonte Bernhard, John Christopher, Andy Elby, Paul Gresser, Apriel Hodari,
and limitations of what is known, I refer you to the references cited in the
Beth Hufnagel, Pratibha Jolly, Bao Lei, Rebecca Lippmann, Laura Lising,
text.
Tim McCaskey, Seth Rosenberg, Mel Sabella, Rachel Scherr, Richard
This book has four parts: Steinberg, Jonathan Tuminaro, Al Sapirstein, Jeff Saul, Zuyuan Wang, and
Michael Wittmann. Through the past decade, my collaborators in the
• An introduction discussing the structure of The Physics Suite and the Activity-Based Physics Group have been invaluable in both helping me
motivation for educational reform in introductory physics. (chapter 1) develop my views on education and in the creation of this book: Pat
• A discussion of what is known about how people think that is relevant Cooney, Karen Cummings, Priscilla Laws, David Sokoloff, and Ron
for physics teaching and learning. (chapters 2 and 3) Thornton.
• Two chapters about assessing individual students learning and
evaluating the success of instruction for a class. (chapters 4 and 5) I would like to thank those people who commented on various versions of
• A survey of various methods for creating learning environments that the text, especially those who helped clarify my descriptions of their work:
can help to improve student learning, including both tips from my own Bob Beichner, Mary Fehrs, Gary Gladding, Ken and Pat Heller, Paula
classroom experience and descriptions of the PER-based curricular Heron, Priscilla Laws, Eric Mazur, Lillian McDermott, Evelyn Patterson,
materials and methods belonging to The Physics Suite and some other David Sokoloff, Ron Thornton, and Maxine Willis. Priscilla Laws and Tim
methods that work well with it. (chapters 6-10) McCaskey did careful readings of my draft and made many valuable
Finally, the book comes with a Resource CD. This contains suggestions.
• our Action Research Kit  a collection of concept and attitude surveys I want to acknowledge a grant from the University of Maryland Graduate
• resources for exploring computer-assisted data acquisition and analysis Research Board that played a major role in allowing me to take sabbatical to
and video data handling write this book. I would also like to thank the Graduate School of
Education at UC Berkeley for hosting that sabbatical with particular thanks
• resources for getting information about PER.
to the following for valuable discussions: Michael Ranney, Andy diSessa,
In the Appendix to this volume, I list the material available on the disk. The
Alan Schoenfeld, and Barbara White. Much of my research that is cited
disk is attached inside the back cover. If it is missing in your copy, contact
here has been supported by the US National Science Foundation and the
John Wiley & Sons to get one.
Fund for the Improvement of Post-Secondary Education of the US
Acknowledgments Department of Education.
Throughout my studies of PER, a number of individuals have been Finally, special thanks are due to my wife, Janice (Ginny) Redish, not only
tremendously helpful, both through their published work and through for her support and encouragement throughout, but for her outstanding skill
personal conversations. First and foremost is Lillian C. McDermott, not and expertise in editing and technical communication. She was immensely
only through her large and informative body of research, but through taking helpful in making the book more readable. She has also been an invaluable
me in as a sabbatical visitor to her well-established research group in PER resource in helping me both find and understand what is known and
at the University of Washington in 1992-93. This gave me an excellent relevant in cognitive science and the study of human behavior.
Teaching Physics with the Physics Suite E. F. Redish

Table of Contents 7/4/2002
2. Modular reasoning structures:

Teaching Physics with the Physics Suite Primitives and facets
3. Activating resources from everyday experience:
Situated cognition
Edward F. Redish Implications of the Cognitive Model for Instruction: Five foothold
principles
TABLE of CONTENTS 1. The constructivism principle
2. The context principle
Preface
3. The change principle
Chapter 1 Introduction and Motivation 4. The individuality principle
5. The social learning principle
Introduction
Some General Instructional Methods Derived from the Cognitive
Typical materials for a physics class
Model
A new alternative: The Physics Suite
Cognitive Conflict
Motivation
Bridging
Who are we teaching and why?
Restricting the frame
The growth of other sciences
Multiple representations
The goals of physics for all
Rethinking the Goals of Physics Instruction
Are we already achieving these goals?0
Extended Content Goals
Figuring out what doesn’t work and what we can do
about it
Introducing Sagredo Chapter 3 There’s More than Content to a Physics Course: The Hidden
Why Physics Education Research? Curriculum
Knowledge as a community map
A Second Cognitive Level
Building the community map for education
Expectations: Controlling Cognition
The impact on teaching of research on teaching and
Expectations about learning
learning
The structure of student expectations: The Hammer
Even good students get the physics blues.
variables
I wouldn’t have believed it if I hadn’t seen it.
Connecting to the real world
Some caveats
Metacognition: Thinking about thinking
What this book is about
Instructional techniques for improving metacognition
Affect: Motivation, Self-image, and Emotion
Chapter 2 Cognitive Principles and Guidelines for Instruction Motivation
Self-image
The Cognitive Model
Emotion
Models of memory
1. Working memory
2. Long-term memory Chapter 4 Extending our Assessments: Homework and Testing
Cognitive resources for learning
Assessment and Evaluation
1. Robust reasoning structures:
Giving Feedback to your Students
Common naïve conceptions
Homework
Teaching Physics with the Physics Suit E. F. Redish

Getting Feedback from your Students Models of the Classroom

Testing The traditional instructor-centered environment
Designing exams The active-engagement student-centered environment
Exams as formative feedback The Population Considered: Calculus-Based Physics
Eight Types of Exam and Homework Questions Characteristics of calculus-based physics students
Multiple-choice and short-answer questions The hidden curriculum and problem solving
Multiple-choice multiple-response questions Some Active-Engagement Student-Centered Curricula
Representation-translation questions
Ranking tasks
Chapter 7 Lecture-Based Methods
Context-based reasoning problems
Estimation problems The Traditional Lecture
Qualitative questions A more interactive approach to the traditional lecture
Essay questions Demonstrations
Peer Instruction / ConcepTests
Interactive Lecture Demonstrations (ILDs)
Chapter 5 Evaluating our Instruction: Surveys
Just-in-Time Teaching (JiTT)
Research-Based Surveys
Why use a research-based survey?
Chapter 8 Recitation and Laboratory-Based Methods
Surveys and the goals of a class
Delivering a survey in your class The Traditional Recitation
Understanding What a Survey Measures: Validity and Reliability A more interactive approach to the traditional recitation
Validity Helping your teaching assistants give better recitations
Reliability Tutorials in Introductory Physics
Content Surveys The structure of Tutorials
The FCI Tutorials often focus on important but subtle points.
The FMCE Should you post solutions to Tutorial pretests and
The MBT homework?
Attitude Surveys What does it take to implement Tutorials?
The MPEX Tutorials produce substantially improved learning gains.
MPEX Results Changing recitations to Tutorials doesn’t hurt problem
Analyzing the MPEX solving.
Getting improvements on the MPEX Students need to get used to Tutorials.
The VASS ABP Tutorials
Scientific dimension of the VASS ABP Tutorials are mathematically and technologically
Cognitive dimensions of the VASS oriented.
The EBAPS Concept learning can be tied to the use of math.
Cooperative-Problem Solving
Cooperative Problem Solving (CPS) relies on context-rich
Chapter 6 Instructional Implications: Some effective teaching methods
problems.
Introduction Group interactions play a critical role.
Research-Based Curricula

The work of the group is better than the work of the Evaluating instruction: The Action Research Kit
best student in it Suite compatible elements:
Techniques for improving group interactions30 Peer Instruction, JITT, and Cooperative
The Traditional Laboratory Problem Solving
Goals of the laboratory Using The Physics Suite in Different Environments6
Often, less actually happens in traditional labs than we The role of room layout
might hope. The role of facilitators
A more interactive approach to the traditional laboratory Four Case Studies: Adopting and Adapting Suite Elements
RealTime Physics Using Suite elements at a small institution
RTP uses cognitive conflict and technology to build Gettysburg High School
concepts. Pacific University
RTP relies on psychological calibration of technology. Using Suite elements at a large institution
RTP labs are effective in building concepts. The University of Illinois
North Carolina State University
Conclusion
Chapter 9 Workshop and Studio Methods
Physics by Inquiry
Bibliography
In PbI, students learn a few topics deeply.
Students may need help in changing their expectations for
PbI.
Evaluations of PbI show it to be very effective.
Workshop Physics
Students in WP build their concepts using technology.
WP is developed through and informed by education
research.
WP changes the frame in which students work.
Evaluations of WP show it to be highly effective in
building concepts.
Chapter 10 Using The Physics Suite

The Idea of The Physics Suite
The Principles Behind The Physics Suite
The Elements of The Physics Suite
The Suite’s narrative text: Understanding Physics
Using the Suite in lab: RealTime Physics
Using the Suite in lecture: Interactive Lecture
Demonstrations
Using the Suite in recitation sections: Tutorials
Putting it all together: Workshop Physics
Homework and exams: Problems and Questions

Appendix (on Resource CD) Reading List for a Graduate Seminar in Teaching College
Physics for Physicists
Sample Problems for Homework and Exams
Reading List for a Graduate Seminar in Physics Education
Estimation Problems
Research (courtesy, Diane Grayson)
Multiple-Choice and Short Answer Problems
Other Resources
Representation Translation Problems
Guidelines and Heuristics: Summary of goals, principles,
Ranking Tasks
and commandments
Open-Ended Reasoning Problems
Writing a Scientific Paper
Context-rich Reasoning Problems
Resources for Computer Assisted Data Acquisition and Analysis
Essay Questions
MBL information from Vernier
JiTT Problems (courtesy, Ellen Patterson)
MBL information from Pasco
Action Research Kit
MBL information from Texas Instruments
The Mathematical Modeling Conceptual Evaluation
Videopoint demonstration
(MMCE)
Information on the Student Response System
The Vector Evaluation Test (VET)
from Classtalk
Test of Understanding Graphics (TUG-K)
WP Excel Tools
Force Concept Inventory (FCI)
Information on the AAPT
Force-Motion Concept Evaluation (FMCE)
The Mechanics Baseline Test (MBT)
Energy Concept Survey (ECS)
Conceptual Survey of Electricity and Magnetism (CSEM)
The Electric Circuits Concept Evaluation (ECCE)
Rate and Potential Test, versions A and B (RAPT)
Wave Diagnostic Test (WDT)
Determining and Interpreting Resistive Electric Circuits
Concept Test (DIRECT)
The Small Particle Model Assessment (SPMA)
The Measurement Uncertainty Quiz (MUQ)
Maryland Physics Expectations Survey (MPEX)
The Views about Science Survey (VASS)
Bibliographic Resources
L. C. McDermott and E. F. Redish, “Resource Letter:
PER-1:
Physics Education Research,” Am. J. Phys. 67, 755-
767 (1999).
L. Jossem, “Resource Letter EPGA-1: The education of
physics graduate assistants,” Am. J. Phys. 68, 502-
512 (2000)
Useful Books: A list of books that contain discussions of
student learning,
innovative teaching methods, and interesting
problems.

CHAPTER 1
Introduction and Motivation
The Master doesn’t talk, he acts.

When his work is done
[his students] say, “Amazing:
we did it, all by ourselves.”
Lao Tse, Tao Te Ching [Mitchell 1988]
INTRODUCTION
Teaching physics can be both inspirational and frustrating. Those of us who enjoy learn-
ing physics get to rethink and pull together what we know in new and coherent ways.
We enjoy the opportunity to create new demonstrations, invent new derivations, and
solve interesting problems. For those of us who love doing physics, teaching can be a de-
lightful learning experience. Occasionally we find a student who has the interest and abil-
ity to understand what we are trying to do and who is inspired and transformed by our
teaching. That makes all the frustrations worthwhile.
On the other hand, there are frustrations. We may have students who seem unable
to make sense of what we do—sometimes a lot of them. They are confused and even
hostile.We may make intense efforts to reach these students, either by making our classes
more entertaining or by simplifying what we ask them to do. While these efforts may
lead to better student evaluations, they rarely result in our students understanding more
physics. They can lead to a “dumbing down” of the physics that we find frustrating and
disappointing.
Can we reduce this frustration and find ways to reach those students who don’t
seem to “get it”? In the past two decades, there has been a growing understanding of
why so many students respond badly to traditional physics instruction and of how to
modify our instructional methods to help them learn more. A number of researchers
and curriculum developers have begun to weave the results of education research and
new technological tools into more effective learning environments.
1
2 • Chapter 1: Introduction and Motivation
One result of this interweaving of research and technology is the Physics Suite. In
the Physics Suite, the Activity-Based Physics (ABP) Group1 is creating a new kind of
educational environment. Since there is a growing diversity of environments for intro-
ductory physics, the ABP Group has opted for a modular structure, one that can be im-
plemented a step at a time or adopted in a total makeover, affecting all parts of the
course. This book is about how our teaching of physics can change as a result of these
new environments. It discusses the elements of this modular structure, how to use them,
and the educational philosophy, cognitive theory, pedagogical research, and modern tech-
nology on which the Physics Suite is based.
TYPICAL MATERIALS FOR A PHYSICS CLASS

Typically, all the materials offered by a publisher for a physics class derive from the text (see
Figure 1.1). Affiliated materials are available associated with the text, usually including every-
thing from a “quick summary” for students to colored transparencies for instructors. There
may be a CD with (often uninteresting) simulations that offer little or no guidance for either
students or instructors in how to use it to make it pedagogically effective. Adopting institu-
tions may add a laboratory with lessons developed on site. But the text is primary, and its se-
lection usually depends critically on content—what is covered and whether it is treated cor-
rectly. Those are certainly important criteria.
But extensive research has shown that effective student learning seldom comes from the
text. Students frequently have difficulty making sense of a physics text, and, as a result, only
a small minority actually read the text in the careful and thoughtful way we expect. Effective
student learning comes from “brains-on” activities—those times when they are thinking hard
and struggling to make sense of what they are learning. Effective instruction happens when
we create environments in which students are encouraged and helped to engage in those kind
of activities. Well-tested innovations that focus on building reasoning through carefully
planned and structured activities in lectures, recitations, laboratories, or workshops are more
Text
Summary book Overheads Simulations Solution book
Lots of unrelated pieces

of uncertain value
Figure 1.1 Typical layout of materials associated with a physics course.
1 Pat Cooney, Karen Cummings, Priscilla Laws, David Sokoloff, Ron Thornton, and myself.
A New Alternative: The Physics Suite • 3
likely to produce strong student learning. For most students, these activities play as impor-
tant a role as does reading the text.
The Physics Suite is much more than a text with a collection of ancillaries developed af-
ter the fact. The Physics Suite builds on integrating a series of strong activity-based elements
with the text. The Physics Suite focuses on getting students to learn to do what they need to
do to learn physics.
A NEW ALTERNATIVE: THE PHYSICS SUITE

The ABP Group has created a new structure consisting of a broad array of integrated educa-
tional materials: The Physics Suite. These materials are shown schematically in Figure 1.2. Two
Problems
Touchstone
examples
Reading
Narrative exercises Action
Research
Problem
Kit
collections
Understanding Concept
Physics surveys
Attitude
surveys
Laboratories ILDs
Instructor’s
RealTime Guide
physics Teaching Physics
Tools for with
Scientific the Physics
Thinking Suite
Tutorials Workshops
Tutorials in Workshop
Introductory Physics
Physics
Tools Explorations
ABP in Physics
WP
Tutorials
spreadsheet
tools
Videopoint
CADAA tools
Figure 1.2 Elements of The Physics Suite

particular themes of many Suite elements are: (1) the use of guided activities to help students
construct their learning, and (2) the use of modern technology, particularly computer-assisted
data acquisition and analysis (CADAA). The Physics Suite consists of the following materials:
• The Instructor’s Guide: Teaching Physics with the Physics Suite (Redish)—This book: a guide not
only to the materials of the Suite (and other curricular materials that fit well with the Suite),
but a discussion of the motivation, theoretical frame, and data describing its effectiveness.
• Narrative: Understanding Physics (Cummings, Laws, Redish, and Cooney)—A revised ver-
sion of the classic Halliday, Resnik, and Walker text, modified to focus more strongly
on issues where students are known to have difficulties from educational research. The
new narrative also stresses the empirical base for physics knowledge, explaining not only
what we know but how we know it. (See chapter 10 for more details on how the text
has been changed.)
• Problems —Since problem solving is one of the places where students learn the most in
a physics class, the Suite is enriched by a careful choice of sample problems in the nar-
rative (Touchstone Problems) and by supplementary problems of a variety of types in-
cluding estimations, representation translations, and context-rich real-world problems.
These are contained in the narrative, on the Resource CD in the back of this book, and
in a supplementary book of problems.
• Action Research Kit—A collection of tools for evaluating the progress of one’s instruction,
including a variety of concept tests and attitude surveys. These are on the Resource CD
that comes with this volume.
• ILDs: Interactive Lecture Demonstrations (Sokoloff and Thornton): Worksheet-based guided
demonstrations that use CADAA to help students build concepts and learn representa-
tion translation.
• Workshops—Three sets of full workshop/studio materials are associated with the Suite.
1. Workshop Physics (Laws): A full lab-based physics course at the calculus level using
CADAA, video, and numerical modeling.
2. Explorations in Physics (Jackson, Laws, and Franklin) : Developed as part of the Work-
shop Science project, a lab-based curriculum that uses less mathematics and is de-
signed for use with nonscience majors and preservice teachers.
3. Physics by Inquiry (McDermott et al.): A workshop-style course appropriate for pre-
and in-service teachers.
• Tools (Laws, Cooney, Thornton, and Sokoloff ) : Computer tools for use in laboratory, tu-
torial, and Workshop Physics include software for collecting, displaying, and analyzing
CADAA data, software for extracting and plotting data from videos, and spreadsheets
for analyzing numerical data.
• Tutorials —Two sets of materials for use in recitation sections that use guided small-group
activities to help build understanding of concepts and qualitative reasoning:
1. Tutorials in Introductory Physics (McDermott et al.)—A collection of worksheets by the
Physics Education Group at the University of Washington (UW).
2. ABP Tutorials (Redish et al.)—Additional tutorials in the UW mode but ones that in-
tegrate the technological tools of the Suite—CADAA, extraction of data from videos,
and simulations. These also extend the range of topics to include modern physics.
Motivation • 5
• Laboratories: A set of laboratories using CADAA to help students build concepts, learn
representation translation, and develop an understanding of the empirical base of physics
knowledge. Two levels of labs belong to the Suite.
1. RealTime Physics (Thornton, Laws, and Sokoloff ): Appropriate for college level physics.
2. Tools for Scientific Thinking (Thornton and Sokoloff ): Similar to RealTime but at a
slower pace for high school physics.
The materials of the Suite can be used independently, but their approach, philosophy,
and notation are coherent. As a result, you can easily adopt one part as a test of the method
when it is convenient and appropriate, or you can integrate several Suite elements, trans-
forming all parts of your class.
Detailed discussions of the various components of the Suite are given in chapters 7–9,
and considerations of how they might be used are presented in chapter 10. Those who are
familiar with the research and motivation behind modern physics education curriculum re-
form are invited to turn to those chapters directly. If you are not familiar with the research
and theory behind these materials, read the rest of this chapter and the next few chapters
where I present some motivation and background.
MOTIVATION
Why do we need the Physics Suite? Most of us learned perfectly well from a text. What is
different today? A number of things have changed and are going to be changing even more
in the future.
• The students we are teaching have changed.

• The goals we want to achieve with these students have changed.
• We know much more today about how students learn than we used to.
• We have more tools to work with—both technology and new learning environments—
than we used to.
I organize my discussion of these points around two questions:
1. Who are we teaching and why?

2. Why Physics Education Research (PER)?
Who are we teaching and why?

Since both the difficulties in teaching physics and their solutions depend on the population
of students we are considering, let’s begin by considering who our students are—and who
they are likely to be in the next few years.
The growth of other sciences

When I began my serious study of science as a high school student more than four decades
ago, it seemed to me that only in physics could I do “real” science. By this, I meant discov-
ering fundamental laws of nature and making sense of their implications. As a high school
student, I was particularly taken by the beautiful match between the mathematics, which I
adored, and the physics, which I took to represent the real world. To a certain extent, I was
right—at least for me. Physics is the crown jewel of the sciences, making an elegant link be-
tween the understanding of the deeply fundamental and the powerfully practical. Einstein’s
“Emc2” and the nuclear bomb that so affected the politics and even the daily sensibilities
of many normal citizens during the last half of the last century are only the tip of the ice-
berg. Quantum mechanics leads us to a deep understanding of the structure of matter, re-
sulting in developments like the transistor and the laser that continue to profoundly change
our way of life.
What I missed as a high school student was the immense progress soon to be achieved
by the other mature sciences, such as biology and chemistry, and the immense growth soon
to be shown in the then infant sciences of computer science and neuroscience, among oth-
ers. Today, a high school student with an interest in science can have exciting opportunities
for a productive and fascinating career in a wide range of sciences, from building models of
the universe to modeling neural processes of the brain. Physics is now only one of many mu-
tually enhancing jewels in the crown of science.
Progress in these other sciences has been facilitated by advances in physics in many
ways, from improvements in gravitational theories to the development of fMRI (functional
magnetic resonance imaging), a tool that uses nuclear magnetic resonance to noninvasively
track changes in brain metabolism as people think about different things. Students going
into these other sciences need to understand physics as part of their scientific education,
but what exactly do they need from us? What role can (and should) physics play as a part
of the education of a professional scientist in biology or chemistry? What role can (and
should) physics play as a part of the education of a technical professional such as an engi-
neer or paramedic?
The goals of physics for all

Physics instruction has traditionally played two very obvious roles in the education of scien-
tists: both to recruit and train professional physicists-to-be and to “filter out” those students
who might not be able to handle the mathematics of engineering or the memorization re-
quired in medical school. The former role becomes a smaller fraction of our teaching activi-
ties as the number of students studying other sciences grows. The latter no longer seems ap-
propriate for present circumstances, when engineers, scientists, and medical professionals have
an increasing need to understand both the systems they are working with and the complex
tools they are using to probe them.
Improving our teaching of physics is more important today than ever before. First, a
larger fraction of the population is graduating from high school and going on to universities
than in previous times. More of these students than ever before are either interested in a sci-
entific career or are concerned about finding jobs in an increasingly technological workplace.
Second, especially for those of us in publicly supported institutions, the governments
and the populace that employ us are more likely today to hold the educational system (and
therefore its teachers and administrators) directly responsible for the students’ learning—or
lack of it—than they were in the past. In other times, individual students were seen to be
personally responsible for their learning and less attention was paid to the effectiveness of
Motivation • 7
teaching.2 Today, workplace demands for more technologically trained personnel require that
we do whatever we can to help facilitate the successful education of our students.
The task of the physics teacher today is to figure out how to help a much larger frac-
tion of the population understand how the world works, how to think logically, and how to
evaluate science. This is doubly important in democratic countries where a significant frac-
tion of the adult population is involved in selecting its leaders—leaders who will make deci-
sions not only on the support of basic science, but on many issues that depend intimately on
technological information. It would be of considerable value to have a large fraction of the
populace who could not be fooled by the misuse of science or by scientific charlatanism.
Are we already achieving these goals?

Does traditional physics teaching “work” in the introductory physics classroom? Unfortu-
nately, the answer seems to be a resounding “no.” Detailed examinations by many physics ed-
ucation researchers have shown that traditional physics instruction does not work well for a
large fraction of our students. Many of our students dislike physics; many feel that it has no
relation to their personal lives or to their long-term goals; and many fail to gain the skills
that permit them to go on to success in advanced science courses.
The nature of the difficulty appears to be a kind of “impedance mismatch.” The
professor sends out information and sees it reflected back in a similar or identical form (Fig-
ure 1.3), but little understanding has actually gotten through to the other side.
Figuring out what doesn’t work and what we can do about it

If we are to improve the situation, the best approach is to use our scientific tools to under-
stand what is going on. We need to observe the phenomena we want to understand and try
to make coherent sense of what we see. From educators and cognitive psychologists, we learn
two important lessons.
• To understand what will work, we have to concentrate on what the student is learning
instead of on what we are teaching.
Professor Student
Professor Student
Figure 1.3 The fact that something “comes back as we sent it out” does not mean that much has
“gotten through to the student,” especially if students possess a large inertia!
2 In the end, every student is indeed responsible for his or her own learning. But the issue is whether students are
to learn everything on their own—no matter what we throw at them—or whether they can learn more with the aid
of appropriately designed learning environments and interactions with trained mentors. This is discussed in more
detail in chapter 2 under the heading, “The Social Learning Principle.”
• We have to do more than evaluate our students’ success. We have to listen and analyze
what they are thinking and how they learn.
If we really want to change how our students think about the physical world, we have to un-
derstand how they think.
Introducing Sagredo
At this point in our discussion, I need to introduce a voice other than my own. Not every-
thing in this book will be obvious to the professional physicist teaching physics, even to one
with years of experience. Much of what has been learned in PER is surprising and counter-
intuitive. Occasionally, contradictory ideas about teaching seem obviously true. In order to
make it easier for you to keep track of both sides of the discussions, I introduce my virtual
colleague, Sagredo.
Sagredo is a successful researcher at a large research-oriented state university. He is dis-
satisfied with what his students learn in his introductory physics classes and has not been able
to improve the situation despite significant and time-consuming efforts. In a grand old physics
tradition [Galileo 1967], I will use Sagredo as a personification of the thoughtful, intelligent
physicist who has little or no experience with physics education research. I’ve chosen Galileo’s
impartial but intelligent listener, Sagredo, since this book is intended for an audience of pro-
fessional physicists and physics teachers, most of whom are highly sophisticated in their knowl-
edge of physics but who may not have thought deeply about the issues of how people think
about physics.
To the last sentence of the previous paragraph, Sagredo might well respond, “I learned
physics with traditional instruction from teachers who didn’t think about how I thought.
Why can’t my students do it the same way as I did?” The reason is that we are not only con-
cerned about training physicists.
I recall well the first time I ever taught electromagnetism (to a class of sophomore physics
majors at Maryland using Purcell’s lovely and insightful text [Purcell 1984]). Suddenly, it
seemed, everything made coherent physical sense. Before that, I knew all the equations and
could even solve lots and lots of Jackson problems [ Jackson 1998] with some alacrity, but I
hadn’t really “made physics of it” as a graduate student—and I hadn’t realized that I hadn’t.
Amused by the different feeling associated with my new knowledge, I realized that I had stud-
ied electromagnetism with Maxwell’s equations five times through my high school, college,
and graduate school years.
“But,” responds Sagredo, “perhaps you needed that. Perhaps one cannot expect someone
to understand physics the first time through.” Certainly one rarely learns something at the
first look. We often have to see something many times before we really learn it. But then we
have a problem. Very few of our students will have the opportunity to do what I did—study
the physics many times at many levels and eventually teach it. We have to decide the value
of one step in a six-step process. If you know your children will take music lessons for 10
years, it might suffice to begin with a year of scales and finger exercises, to strengthen their
hands. But it might not. Many, perhaps most, children will rebel and not even make it to a
second year.
So the hard question for us is: Is it possible to provide some useful understanding of
physics to students in a one-year physics course? Or does the real utility of the course only
Why Physics Education Research? • 9
come in providing a foundation for future physics courses? If the latter is true, few of the
students who are now taking introductory high school or university physics would be well
advised to continue their efforts, since most will not take any future physics courses. Fortu-
nately, as we begin to understand how students learn physics, we begin to see that remark-
able improvements in understanding are possible for a large fraction of students in our in-
troductory courses.
At this point, Sagredo complains: “But if you modify introductory physics so that the
average student does better, aren’t you going to bore the very best students? These students
are important to us! They are the future physicists who will be our graduate students and our
successors.” Sagredo, I agree that we would need to be concerned if that were the case. If we
were to improve our instruction to the middle 50% while degrading it for the top 5%, it
could be a disaster for the profession. What is particularly gratifying is that the improved
learning that takes place as a result of instructional reform based on an understanding of how
students think is not limited to the “middle of the road” student who was previously getting
by but was not getting much of long-lasting value from the course. Over the past decade,
physics education research has consistently documented that the top students in the class
show even stronger gains than the midrange students from research-motivated cognitive-based
curriculum reforms. (See, for example, [Cummings 1999].)
WHY PHYSICS EDUCATION RESEARCH?3

The ABP Group brings together individual physics education researchers and curriculum de-
velopers who are working as a part of a community effort to improve both our understand-
ing and our implementation of physics teaching. An important component of this effort is
the word “community.” We share the philosophy with a growing cadre of physicists that to
teach physics more effectively, we need to work together as a research and development com-
munity. We need to work together in the way that scientists in a science and industry sector
work together to improve both our knowledge of how the world works and of how to
make use of that knowledge. We share the conviction that by using the tools of science—
observation, analysis, and synthesis—we can better understand how students learn and find
ways to improve how we teach as a result.
Why do we need to go beyond the usual observations we have made, both of our teach-
ers when we were students and of our students now that we are the teachers? To answer this
question, we need to think about the nature of our knowledge—both of science and of
teaching.
To understand how people learn science and how we might use science to learn about
how people learn, we need to think a bit about the nature of the knowledge we are learning.
We often say that the goal of science is to discover the laws of nature. This is not quite pre-
cise enough for our purposes. It’s better to say that we are a community working together to
create the best way of thinking about the world that we can. This places the knowledge firmly
where it really resides—in the head of the scientist as a part of a scientific community.
3 Much of this section is based on my Millikan lecture [Redish 1999].

Knowledge as a community map

A good metaphor for the process of science is the building of a map. A map of the world
should not be mistaken for the world,4 but it can nonetheless be of great value in getting
around. What is perhaps most important about the scientific map of the world is that it is
more than just the collection of the maps of individual scientists. The culture of science in-
cludes the continual interaction, exchange, evaluation, and criticism we make of each others’
views. This produces a kind of emergent phenomenon I refer to as a community consensus
knowledge base, or more briefly, a community map. I visualize this as an idealized atlas of sci-
ence. Just as an atlas contains many individual charts, so the atlas of science contains many
distinct, coherent, but incomplete areas of knowledge. These areas are supposed to agree where
they overlap, but it is not clear that the entire universe can be encompassed in a single map.5
No single individual, no matter how brilliant, has a map identical to this community con-
sensus map.6
At this point, Sagredo might again complain. “But science isn’t just the collection of in-
dividual scientists’ knowledge. What we know in physics is the correct description of the real
world.” Sagredo, I agree. But we have to be a bit more explicit about what this “correct”
knowledge is and where it resides. If no one individual has the complete map, but all knowl-
edge ultimately lies in someone’s head, in what sense does the knowledge of the world we
have gained as a community exist? The key is in the phrase “as a community.”
Real maps are constructed in a manner similar to the way we construct science. They
are built by many surveyors. No one surveyor has made all the measurements that lead to a
map of the United States, for example. Furthermore, each atlas differs in some detail from
every other atlas, yet we have little doubt that a true atlas could exist (though it would, of
course, have to be dynamic and limited to a preset resolution).
In mathematics, if we have a series of functions that get closer and closer to each other
in a prescribed way, then we say the sequence has the Cauchy property.7 Even if we can’t find
the true limit analytically, we find it convenient to act as if such a limit exists.8 The natural
mathematical structures of sets of functions behave much more nicely if we “complete” the
space by adding the sets of Cauchy sequences to our space. It’s like adding the real numbers
that fall in between the rationals. We can never calculate them exactly, but it would be very
hard to describe the phenomenon of motion if we left them out. (See Figure 1.4.)
In many areas of physics, the sequence of knowledge functions has converged—for all
practical purposes. The community consensus on such items as the classical mechanics of the
planets of the solar system or the thermodynamics of weakly interacting gases, for example,
4 Lewis Carroll describes a community of mapmakers who are creating increasingly accurate maps. Finally, they cre-
ate a map of the area that is 1-1 in scale. Unfortunately, the local inhabitants refuse to let them unroll it “because
it will block the sun and kill the crops” [Carroll 1976, p. 265].
5 Mathematically, this is even true of a sphere, which cannot be mapped by a single non-singular map to a Euclid-
ean plane. See, for example, [Flanders 1963].

6 In some areas, a specific individual’s map may be better than the community’s map.
7 Mathematically stated, a sequence of functions {f (x)} is said to be Cauchy if two functions taken from far enough
n
out in the sequence will be as close together everywhere as you want. (Given any 0 there is an N such that if
m,n N, fn(x) fm(x) for all x.)
8 For mathematical details, see for example, [Reed 1980, p. 7].
An emergent phenomenon— Community

a Cauchy sequence consensus
knowledge
A convenient
and useful
representation
Individual
scientists’ Physical
As a result of
knowledge world
training and
experience
Figure 1.4 Representation of the process of building the scientific map of the physical world.
is exceedingly strong—in part because we know the resolution that is relevant to most prob-
lems in these subjects. Just as we don’t find it useful to have a map of New York that speci-
fies the cracks in the sidewalk, we don’t need to calculate the location of a satellite to nanome-
ter accuracy.
Building the community map for education

If what we learn about physics education is to lead to a stable and growing community map,
the community needs to document what we know and to present conjectures and hypothe-
ses for criticisms and questioning. This is particularly important in education.
Human behavior in all realms is beset by wishful thinking—we tend to really believe
that what we want to be true is true. To some extent, the most important part of the process
by which science builds its community consensus knowledge base is the part that identifies
and purges the wishful thinking of individual scientists. Some parts of the process critical for
this task include:
• Publication of results, documented with sufficient care and completeness that others can
evaluate and duplicate them
• Repetition of experiments using different apparatus and different contexts9
• Evaluation and critiquing of one scientist’s results by others through refereeing, presen-
tations and discussions in conferences, and through follow-up evaluations and extensions
When it comes to education, wishful thinking is not just present; it is widespread and can
take a variety of forms.
• A dedicated and charismatic teacher may, by force of personality, inspire her students
into learning far above the norm. That teacher may then try to disseminate her cur-
riculum to other less charismatic individuals, only to find the method is no longer
effective.
9 Wetry to make experiments as similar as possible, but it is not, of course, possible to ever reproduce an experi-
ment exactly—even if the identical apparatus is used. These small variations help us understand what variables are
important (e.g., the colored stripes on the resistors) and which are not (e.g., the color of the insulation on the wires).
• A teacher delivering an inappropriately rigorous course may find that his students seem
to learn little and to dislike it intensely. “Ah,” he is heard to remark, “but when they’re
older they will realize that I was right and come to appreciate the course and what they’ve
learned.”
• A teacher concerned about how little his students are learning may try a number of
changes to improve the situation, but find that nothing seems to help. “Oh well,” he
says, “those students are just not able to learn physics under any circumstances.”10
I have personally observed each of these responses from physics colleagues whose science and
whose teaching efforts I respect. Each of these situations is more complex than the individ-
ual teacher has realized. In each of these situations, much more can be done if a deeper un-
derstanding of learning and teaching is brought to bear.
Building a community knowledge base about education requires using our full array of
scientific tools—observation, analysis, synthesis, plus the community purging and cleaning
tools of publication, repetition, and evaluation. Sagredo complains at this point. “Do we re-
ally need all this effort? Oh, I know you’re right about some teachers, and I recognize your
three ‘wishful thinkings’—I’ve occasionally been there myself. But there are some good teach-
ers. I’ve had one or two. Why don’t we just let them concentrate on the teaching and carry
most of the introductory load?”
Yes, Sagredo, there have always been excellent teachers—teachers who can reach not only
the very best students, but who can energize and educate even the less motivated and less ca-
pable students. But there are far too few of them, and their skill has not been easily trans-
ferable to other concerned and dedicated, but less naturally talented, teachers. We want to
understand what it is those teachers are doing successfully so as to be able to transform suc-
cessful teaching from an art, practiced only by a few unusual experts, to a technology that
can be taught, learned, and facilitated by powerful tools. Building an understanding of the
educational process through using the tools of science is beginning to enable us to carry out
this transformation.
The impact on teaching of research on teaching and learning

Sagredo is still not convinced. “How can education research in someone else’s class tell me
anything about what is happening in my own? Every situation is different—different stu-
dents, different teacher, different university. Each of those differences is important.” You are
right, Sagredo, there are differences and they matter. But a lot of what has been learned is
robust enough to illuminate what is happening in many different situations. Let me illustrate
this with specific examples of how research has affected two instructors.
Even good students get the physics blues

An example of how building a scientific community of physics education researchers
can spread and transform individuals is told by Eric Mazur, a chaired professor at Harvard
10 Note from this example that wishful thinking does not necessarily imply a rosy view of a situation. It may be that
the wishful thinking is that “the situation is so bad that there is nothing I can do about it and therefore I don’t have
to make an effort.”
1. In the circuit at the right, find the current 8V

through the 2 ohm resistor and the potential 1Ω
difference between point a and b.
b
2Ω
a
12 V
1Ω
2. In the circuit at the left, explain what will

happen to the following variables when the
switch is closed:
• the current through the battery
• the brightness of the bulbs
• the voltage drop across the bulbs
• the total power dissipated.
Figure 1.5 A quantitative and a qualitative problem in DC circuits (from [Mazur 1997]).
University. Mazur read the paper published by Ibrahim Halloun and David Hestenes in 1985
in which they described common student conceptual difficulties revealed by physics educa-
tion research [Halloun 1985a]. Mazur was quite skeptical, being reasonably satisfied with the
grades his students achieved on his examinations. Halloun and Hestenes had included a sur-
vey instrument in their paper, a 29-item multiple-choice test probing students’ understand-
ing of fundamental concepts in mechanics.11 Mazur looked at the questions and found them
trivial. He was certain his Harvard students would have no trouble with any of the questions.
He decided to give the test to his introductory physics class after appropriate instruction
[Mazur 1992]. Upon looking at the questions, one student asked: “Professor Mazur, how
should I answer these questions? According to what you taught us, or by the way I think
about these things?”12 Mazur was appalled at how many “trivial” questions his students missed.
(See, for example, the problem shown in Figure 4.1.) He began to look at his teaching as a
research problem.
Mazur went on to study and document in detail the difference between the algorithmic
problem-solving skills his students displayed and the conceptual understanding that he had
been assuming automatically came along for the ride [Mazur 1997]. On an examination to
his algebra-based physics class, he gave the problems shown in Figure 1.5.
11 Anupdated version of this exam, the Force Concept Inventory (FCI), is on the Resource CD distributed with this
volume. Also see the discussion of the FCI in chapter 5.
12 This confirms both Mazur’s and my prejudice that Harvard students are good students. Many students have this
same dichotomy but are not aware that it exists. The Harvard students actually did well on the exam—given our
current understanding of what can be expected on this exam after traditional instruction—just not as well as Mazur
expected.
The average score on the first problem was 75%; the average score on the second was
40%. Students found problem 2 much more difficult than problem 1, despite the fact that
most physicists would consider the analysis of the second problem, the short circuit, much
simpler; indeed, parts of it might be considered trivial. This study and many others13 show
that students frequently can solve complex algorithmic problems without having a good un-
derstanding of the physics.
Before his epiphany, Mazur was a popular and entertaining lecturer. After his encounter
with physics education research, he became a superb and effective teacher.
I wouldn’t have believed it if I hadn’t seen it

I can provide a second example of the value of the community exchange of research results
from my personal experience. When Arnold Arons’ book on teaching physics [Arons 1990]
first appeared, I was absolutely delighted. Although I was not yet a physics education re-
searcher, I had had a strong interest in physics teaching for many years. I had read many of
Arons’ papers and had great respect for them. I read the book cover to cover and annotated
it heavily. In chapter 6 (p. 152) you will find the sentence: “This paves the way for elimi-
nating misconceptions such as repulsion between a north magnet pole and a positive electric
charge, and so on.” I wasn’t very worried about this. It isn’t even underlined in my copy of
Arons. (I underlined about a fifth of the sentences in that chapter.)
But in January of 1994, the Physics Education Group (PEG) at the University of Wash-
ington reported the results of a study of engineering students’ responses to being taught about
magnets [Krause 1995]. Traditionally, many teachers and textbook writers assume, just as I
did, that students know little about the subject, so a good way to introduce it is by analogy
with electric charge, the topic typically presented just before magnetism. The Washington
PEG demonstrated that before the lectures on magnetism, more than 80% of their engi-
neering students confused electric charges and magnetic poles as measured by the simple prob-
lem shown in Figure 1.6. After traditional instruction, this number remained above 50%. I
was both flabbergasted and distressed at hearing this. I had taught the subject off and on for
nearly 25 years and was teaching it at the time of the presentation. Furthermore, I believed
that I listened carefully to students, and I was already sensitized to the issue that students
bring their previous knowledge to any new learning experience. Yet I had never imagined
such a confusion was common. I probed my class upon my return and, needless to say, found
exactly the same results as the Washington group.
The Arons book is still one of the best “teacher-to-teacher” books available. But despite
my respect for Arons’ insights, I was skeptical about the importance of a possible student
confusion between electric charge and magnetic poles. Indeed, I felt my personal experience
contradicted it. The point was only convincingly brought home to me by the solid experi-
mental data offered by the UWPEG.14
13 Forexample, [Halloun 1985b].

14 Note further that this result had been known previously and even published, but not in a journal that I looked at
regularly or that was conveniently available. See [Maloney 1985].
Some Caveats • 15
A bar magnet is hung from a string through its center.

N S
A charged rod is slowly brought up as shown. In what
direction will the magnet tend to rotate?
+
(The magnet will not rotate since electric charges and magnetic poles −
do not exert static forces on one another.)
Figure 1.6 Problem that reveals student confusions about electric and magnetic poles.
SOME CAVEATS
Education research deals with an extremely complex system. At present, neither the educa-
tional phenomenology growing out of observations of student behavior nor the cognitive sci-
ence growing out of observations of individual responses in highly controlled (and sometimes
contrived) experiments has led to a single consistent theoretical framework. Indeed, it is some-
times hard to know what to infer from some particular detailed experimental results.
Yet those of us in physics know well that advancement in science is a continual dance
between the partners of theory and experiment, first one leading, then the other. It is not
sufficient to collect data into a “wizard’s book” of everything that happens. That’s not sci-
ence. Neither is it science to spout high-blown theories untainted by “reality checks.” Science
must build a clear and coherent picture of what is happening at the same time as it contin-
ually confirms and calibrates that picture against the real world.
At present, the combination of education research, cognitive research, and neuroscience
does not provide us with a consistent or coherent picture of how students’ minds function
when they learn physics. Indeed, many problems have been caused by inappropriately
generalizing observations or by interpreting some tentative psychological theory as hard
and fast. Using rules generated by behavioral research incautiously without reference to
continual checks against experimental data (in the classroom!) can lead us to the wrong
conclusions.
But in many cases, educational research has been able to tell us what does not work. And
although it does not provide prescriptive solutions, I have found that the results of educa-
tional research and cognitive science help me to organize my thinking about my students and
to refocus my attention. Instead of concentrating only on the organization of the physics con-
tent, I now also pay attention to what my students are doing when they interact with a physics
course. This is not to suggest that an emphasis on content is somehow unimportant or should
be neglected. What we are teaching is important, but it must be viewed in the context of
how our students learn.
WHAT THIS BOOK IS ABOUT

In this book, my goal is to provide a guide for teachers of physics who are interested in im-
plementing some of the best modern methods that have been developed as a result of the
community’s taking a scientific approach to figuring out how to teach physics. The elements
of the Physics Suite (as well as some others that match well and are easily integrated into a
course using Suite elements) are discussed in chapters 6–10.
It is important to realize, however, that although excellent student-centered approaches
to teaching physics have been developed, none of them are “plug-and-play.” Student-centered
instruction doesn’t mean students are left on their own to do whatever they choose. These
modern approaches require that instructors provide their students with substantial guidance
and learn to work with their students in new ways. That requires that the instructor be rea-
sonably well informed about the premises and methods that are being used.
Most of the literature that backs up these new methods discusses student difficulties to-
gether with explicit data documenting the frequencies and environments in which those dif-
ficulties occur, when they are known. (See the Resource Letter on Physics Education Research
on the Resource CD distributed with this volume [McDermott 1999].)
As a theoretical physicist, I am uncomfortable with providing masses of data without
trying to put them into a theoretical frame. Having such a frame helps make sense both of
what is seen in the research literature and what is seen in the classroom. The appropriate
frame for making sense of educational data is an understanding of how students (and people
in general) think and reason. Therefore, I include a fairly extensive chapter on the relevant
elements of cognitive science and their implication for instruction (chapter 2). This may seem
strange as a component of what is essentially a physics book, but teaching physics is not only
about physics: it is also about how we think about physics.
So it would be useful to have some understanding of how our students think. Sagredo
is nervous about this. “I studied psychology in college. It was all about silly, unrepeatable
things like dreams or irrelevant things like rats running mazes. There were all kinds of con-
flicting schools and fads that came and went. Is there anything really useful there for us?”
Yes, Sagredo, I too took psychology in college (in 1960) and came away mostly disappointed,
but much has happened since then. There are still schools and conflicting opinions, but in-
terestingly enough, there is beginning to emerge a broad consensus, at least on a number of
elements that can be useful for us.
In the next two chapters, I discuss those elements of a cognitive model of thinking and
learning that are relevant for physics education, and I give guidelines and heuristics that can
help us better understand and improve our teaching.
CHAPTER 2
Cognitive Principles and

Guidelines for Instruction1
He who loves practice without theory

is like the sailor who boards ship
without a rudder and compass
and never knows where he may cast.
Leonardo da Vinci
quoted in [Fripp 2000]
When we present instruction to our students, we always build in our assumptions: our ex-
pectations as to what students will do with whatever we give them, our assumptions about
the nature of learning, and our assumptions about the goals of our particular instruction.
Sometimes those assumptions are explicit, but more frequently they are unstated and rarely
discussed. Some pertain to choices we get to make, such as the goals of instruction. Oth-
ers are assumptions about the nature and response of a physical system—the student—
and these are places where we can be right or wrong about how the system works.
If we design our instruction on the basis of incorrect assumptions about our stu-
dents, we can get results that differ dramatically from what we expect. To design effec-
tive instruction—indeed to help us understand what effective instruction means—we
need to understand a bit about how the student mind functions. Much has been learned
about how the mind works from studies in cognitive science, neuroscience, and educa-
tion over the past 50 years. In this chapter and the next, I summarize the critical points
of the cognitive model and organize the information in a way that relates to the in-
structional context. I then consider some specific implications for physics instruction: the
impact of considering students’ prior knowledge and the relevant components of physics
learning other than content. The chapter ends with a discussion of how our explicit cog-
nitive model of student learning can provide guidelines to help us both understand what
is happening in our classroom and improve our instruction.
1 This chapter is based in part on the paper [Redish 1994].
17
18 • Chapter 2: Cognitive Principles and Guidelines for Instruction
THE COGNITIVE MODEL

To understand learning, we must understand memory—how information is stored in the
brain. Modern cognitive science now has complex and detailed structural information about
how memory works. For some simple organisms like the marine snail Aplysia,2 the process is
understood down to the level of neuron chemistry [Squire 1999]. We don’t need that level
of detail for the “application” of understanding physics teaching and learning. A few simple
principles will help us understand the critical issues.
Models of memory
It is clear, from all the different things that people can do that require memory, that mem-
ory is a highly complex and structured phenomenon. Fortunately, we only need to under-
stand a small part of the structure to get started in learning more about how to teach physics
effectively. There are a few critical ideas that are relevant for us. First, memory can be divided
into two primary components: working memory and long-term memory.
• Working memory is fast but limited. It can only handle a small number of data blocks,
and the content tends to fade after a few seconds.
• Long-term memory can hold a huge amount of information—facts, data, and rules for
how to use and process them—and the information can be maintained for long periods
(for years or even decades).
• Most information in long-term memory is not immediately accessible. Using information
from long-term memory requires that it be activated (brought into working memory).
• Activation of information in long-term memory is productive (created on the spot from
small, stable parts) and associative (activating an element leads to activation of other
elements).
In the rest of this section, I elaborate on and discuss some of the characteristics of mem-
ory that are particularly relevant for us.
1. Working memory
Working memory appears to be the part of our memory that we use for problem solving,
processing information, and maintaining information in our consciousness. Cognitive and
neuroscientists have studied working memory rather extensively. Not only is it very impor-
tant to understand working memory in order to understand thinking, but working memory
can be studied with direct, carefully controlled experiments [Baddeley 1998]. For our con-
cerns here, two characteristics are particularly important:
• Working memory is limited.

• Working memory contains distinct verbal and visual parts.
2 Aplysiahas a nervous system with a very small number of neurons—about 20,000, some of them very large—and
a very simple behavioral repertoire. As a consequence, it is a favorite subject for reductionist neuroscientists. See, for
example, [Squire 1999].
The Cognitive Model • 19
Working memory is limited. The first critical point about working memory for us to con-
sider is that working memory can only handle a fairly small number of “units” or “chunks”
at one time. Early experiments [Miller 1956] suggested that the number was “7 2”. We
cannot understand that number until we ask, “What do they mean by a unit?” Miller’s ex-
periments involved strings of numbers, letters, or words. But clearly people can construct very
large arguments! If I had to write out everything that is contained in the proof of a theorem
in string theory, it would take hundreds, perhaps thousands, of pages. The key, of course, is
that we don’t (write out everything, that is). Our knowledge is combined into hierarchies
of blocks (or chunks) that we can work with even with our limited short-term processing
ability.
You can see the structure of working memory in your own head by trying to memorize
the following string of numbers:
3 5 2 9 7 4 3 1 0 4 8 5
Look at it, read it aloud to yourself, or have someone read it aloud to you; look away for 10
seconds and try to write the string down without looking at it. How did you do? Most peo-
ple given this task will get some right at the beginning, some right at the end, and do very
badly in the middle. Now try the same task with the following string
1 7 7 6 1 8 6 5 1 9 4 1
If you are an American and if you noticed the pattern (try grouping the numbers in blocks
of four), you are likely to have no trouble getting them all correct—even a week later.
The groups of four numbers in the second string are “chunks”—each string of four num-
bers is associated with a year, and is not seen as four independent numbers. The interesting
thing to note here is that some people look at the second string of numbers and do not au-
tomatically notice that it conveniently groups into dates. These people have just as much
trouble with the second string as with the first—until the chunking is pointed out to them.
This points out a number of interesting issues about working memory.3
• Working memory has a limited size, but it can work with chunks that can have consid-
erable structure.
• Working memory does not function independently of long-term memory. The interpre-
tation and understanding of items in working memory depend on their presence and as-
sociations in long-term memory.
• The effective number of chunks a piece of information takes up in working memory de-
pends on the individual’s knowledge and mental state (i.e., whether the knowledge has
been activated).
This second point is fairly obvious when we think about reading. We see text in terms of
words, not in terms of letters, and the meanings of those words must be in long-term
3 Of course in this example, it is not simply the chunking that makes the numbers easy to recall. It is the strong
linkage with other, semantic knowledge.
storage. The third point is something we will encounter again and again in different contexts:
How students respond to a piece of information presented to them depends both on what
they know already and on their mental state—what information they are cued to access.
The number of chunks a piece of information has for an individual depends not only
on whether or not they have relevant associations but on how strong and easily accessible that
knowledge is in long-term memory. When a bit of knowledge—a fact or process—is easily
available and can easily be used as a single unit in working memory, we say the knowledge
is compiled. Computer programming is a reasonably good metaphor for this. When code in
a high-level computer language has to be translated line by line into machine instructions,
the code runs slowly. If the code is compiled directly so that only machine-language instruc-
tions are presented, the code runs much more quickly.
Some of the difficulties students encounter—and that we encounter in understanding
their difficulties—arise from this situation. Physics instructors work with many large blocks of
compiled knowledge. As a result, many arguments that seem simple to them go beyond the
bounds of working memory for their students. If the students have not compiled the knowl-
edge, an argument that the instructor can do in a few operations in working memory may re-
quire the student to carry out a long series of manipulations, putting some intermediate in-
formation out to temporary storage in order to carry out other parts of the reasoning.
Studies with subjects trying to recall strings of information indicate that items fade from
working memory in a few seconds if the subject does not try to remember the information by
repeating it consciously [Ellis 1993]. This working memory repetition is known as rehearsal.
Think about looking up a telephone number in a phonebook. Most of us can’t remember it—
even for the few seconds needed to tap in the number—without actively repeating it.
The short lifetime of working memory has serious implications for the way we com-
municate with other people, both in speaking and in writing. In computer science, the hold-
ing of information aside in preparation for using it later is called buffering, and the storage
space in which the information is placed is called a buffer. Since human memory buffers are
volatile and have a lifetime of only a few seconds, it can be very confusing to present infor-
mation that relies on information that has not yet been provided. Doing this can interfere
with a student’s ability to make sense out of a lecture or with a websurfer’s ability to under-
stand a webpage.4
Working memory contains distinct verbal and visual parts. A second characteristic of
working memory that has been well documented in the cognitive literature5 is that working
memory contains distinct components. At least the verbal component (the phonological loop)
and the visual component (the visual sketchpad) of working memory appear to be distinct. (I
am not aware of any evidence in the cognitive literature about the independence of other
components such as mathematical or musical.) This has been demonstrated by showing that
two verbal tasks or two spatial tasks interfere with each other substantially more than a vi-
sual task interferes with a verbal task, or vice versa.6
4 In the theory of communications, this leads to the “given-new principle” in conversation [Clark 1975] and writ-
ing [RedishJ 1993].

5 See [Baddeley 1998] and [Smith 1999] and the references therein.
6 The evidence for this is also strong from neurophysiological lesion studies [Shallice 1988].
2. Long-term memory
Long-term memory is involved in essentially all of our cognitive experiences—everything from
recognizing familiar objects to making sense of what we read. An important result is that re-
call from long-term memory is productive and context dependent.
Long-term memory is productive. What we mean here by “productive” is that memory re-
sponse is active. Information is brought out of long-term storage into working memory and
processed. In most cases, the response is not to simply find a match to an existing bit of data,
but to build a response by using stored information in new and productive ways. This con-
struction is an active, but in most cases, an automatic and unconscious process. Think of lan-
guage learning by a small child as a prototypical example. Children create their own gram-
mars from what they hear.7 Another model of the recall process is computer code. A result,
such as sin(0.23 rad), may be stored as tables of data from which one can interpolate or as
strings of code that upon execution will produce the appropriate data. Analogs of both meth-
ods appear to be used in the brain.
Another example demonstrates that it’s not just sensory data that the brain is process-
ing; cognitive processes such as recall and identification of objects is also productive. Look at
the picture in Figure 2.1. It consists of a number of black blobs on a white background. The
subject of the picture will be immediately obvious to some, more difficult to see for others.
(See the footnote if you have looked at the picture for a while and can’t make it out.8) Even
though you may never have seen the particular photograph from which this picture was con-
structed, your mind creates recognition by pulling together the loosely related spots and “con-
structing” the image. Once you have seen it, it will be hard to remember what it looked like
Figure 2.1 A picture of an animal [Frisby 1980].
7 The fact that they don’t always create the same rules as their parents have is one of the facts that causes languages
to evolve.
8 The picture is of a Dalmatian dog with head down and to the left, drinking from a puddle in the road, seen through
the shadows under a leafy tree.

to you when you couldn’t see it. When you couldn’t “see” the picture in the blobs, was there
a picture there? Now that you see it, where is the picture, on the paper or in your brain?
Long-term memory is context dependent. By context-dependent, I mean that the cognitive

response to a mental stimulus depends on both (1) the external situation and the way in which
the stimulus is presented and (2) the state of the respondent’s mind when the stimulus is pre-
sented. The first point means, for example, that for a problem on projectile motion presented
to a student in a physics class, a student might bring out of long-term memory a different
repertoire of tools than the ones she might access if the same problem arose on the softball
field. To see what the second point means, consider for example, a situation in which a stu-
dent is asked to solve a physics problem that could be solved using either energy or force meth-
ods. If the problem is preceded by a question about forces, the student is much more likely
to respond to the problem using forces than if the question were not asked [Sabella 1999].
To show how the context dependence affects the resources that one brings to the anal-
ysis of a situation, let’s look at the following example.9 Suppose I am holding a deck of
3 5 file cards. Each card has a letter on one side and a number on the other. I remove four
cards from the deck and lay them on the table as shown in Figure 2.2.
I make the following claim: This set of four cards satisfies the property that if there is a
vowel on one side of the card, then there is an odd number on the other. How many cards do
you need to turn over to be absolutely certain that the cards have been correctly chosen to
satisfy this property?
Try to solve this problem before looking to the footnote for the answer.10 Careful! You
have to read both the claim and the question carefully. A similar problem but with a differ-
ent context is the following.
You are serving as the chaperone and bouncer11 at a local student bar and coffee house.
Rather than standing at the door checking IDs all the time, you have occupied a table so you
can do some work. When patrons come in and give their order, the servers bring you cards
K 2 A 7
Figure 2.2 An abstract problem. (See text.)
9 Adapted from [Wason 1966] and [Dennett 1995].

10 You may have to turn over at most two cards to be sure I am telling the truth. (If the first card fails, you know I
am wrong right away.) The only cards that are relevant are the number “2” and the letter “A.” Note that the state-
ment only says “if,” not “if and only if.” To test whether p q, you have to test the equivalent statements: p q
and q p.
11 This is a highly culture-dependent example. In order to solve it, you must know that in most American commu-
nities, the law prohibits the purchase of alcoholic beverages by individuals younger than 21 years of age. Putting the
problem into this cultural context also broadens the concerns of some respondents as well as their tools for solving
it. Some worry whether or not the servers can be trusted or might be lying if, for example, a friend were involved.
Gin
16 Coke 52 &
Tonic
Figure 2.3 A more concrete card problem. (See text.)
with the patron’s order on one side and their best guess of the patron’s age on the other. You
then decide whether to go and check IDs. (The servers can be assumed to be trustworthy
and are pretty good guessers.)
A server drops four cards on the table. They land as shown in Figure 2.3. Which cards
would you turn over in order to decide whether to go back to the table to check IDs?12
This problem is mathematically isomorphic to the previous one, yet most American adults
find the first problem extremely difficult. They reason incorrectly or read the problem in an
alternative way that is plausible if a word or two is ignored or misinterpreted. I have pre-
sented these problems in lectures to physics departments many times. More than two-thirds
of physicists produce the wrong answer to the K2A7 problem after a minute or two of con-
sideration. Almost everyone is able to solve the second problem instantly.
These problems provide a very nice example of both productive reasoning and context
dependence. In the two cases, most people call on different kinds of reasoning to answer the
two problems. The second relies on matching with social experience—a kind of knowledge
handled in a much different way from mathematical reasoning.
This result has powerful implications for our attempts to instruct untrained students in
physics. First, it demonstrates that the assumption that “once a student has learned some-
thing, they’ll have it” is not correct. Most physics instructors who have tried to use results
the students are supposed to have learned in a math class are aware of this problem. The ex-
ample shows that even changing the context of a problem may make it much more difficult.
Second, it points out that a problem or reasoning that has become sufficiently familiar to us
to feel like 16/Coke/52/Gin & tonic may feel like K2A7 to our students! We need to main-
tain our patience and sympathy when our students can’t see a line of reasoning that appears
trivial to us.
Long-term memory is structured and associative. The examples in the previous subsec-
tions illustrate the fundamental principle that the activation of knowledge in long-term
memory is structured and associative. When a stimulus is presented, a variety of elements of
knowledge may be activated (made accessible to working memory). The particular elements
that are activated can depend on the way the stimulus is presented and on the state of the
mental system at the time (the context). Each activation may lead to another in a chain of
spreading activation [Anderson 1999].
12 You would only have to turn over the cards labeled “16” and “Gin & Tonic.” You are not concerned with what
a person clearly much older than 21 orders, and anyone is allowed to order a coke.
Figure 2.4 An example illustrating linked context-dependent schema.
The key to understanding student reasoning is understanding the patterns of association

that activate knowledge elements. In general, a pattern of association of knowledge elements
is sometimes referred to as a knowledge structure. A pattern that tends to activate together with
a high probability in a variety of contexts is often referred to as a schema (plural, schemas or
schemata) [Bartlett 1932] [Rumelhart 1975]. This is illustrated in Figure 2.4. Each circle rep-
resents an element of knowledge. Arrows indicate a probability that the activation of one el-
ement will lead to the activation of another. The different colorings of the circles indicate
schemas—associated knowledge elements that tend to be activated together. Notice that on
the left, some elements have multiple circles. This indicates that particular knowledge ele-
ments can activate different schemas, depending on context.
As an example, consider meeting a new person at a beach party. In your conversation
with this individual, you activate a number of responses—seeking knowledge of your own
about topics the other person raises, looking for body language that demonstrates interest in
continuing the conversation, and so on. If, at some point later in the party, the individual
falls and is knocked unconscious, a different chain of knowledge and responses is activated.
Is the person seriously injured? Do you need to get them to an emergency room? Should the
person be moved or medical personnel called? You still begin with a response to the individ-
ual as a person, but different associated knowledge elements are activated.13
When a schema is robust and reasonably coherent I describe it with the term mental
model. Since scientific models tend to be organized around the existence, properties, and in-
teraction of objects, when a mental model has this character I refer to it as a physical model.
A physical model may or may not agree with our current community consensus view of
physics. For example, the phlogiston picture of thermodynamics was organized around an
13 In reading this situation, did you envision the person as an individual as the same or opposite sex to your own?
This component of context is an example of context affecting activation responses.

image of a physical substance that we now understand cannot be made consistent with phys-
ical observations, so this physical model does not agree with our current community physi-
cal model. Two individuals well versed in a topic may have different physical models of the
same system. For example, a circuit designer may model electric circuits in terms of resistors,
inductors, capacitors, and power sources—macroscopic objects with particular properties and
behaviors. A condensed-matter physicist may think of the same system as emerging from a
microscopic model of electrons and ions.
Cognitive resources for learning

The fact that the mind works by context-dependent patterns of association suggests that stu-
dents reason about physics problems using what they think they know by generalizing their
personal experience. This doesn’t sound surprising at first, but we may be surprised at some
of its implications.
When I learned14 that students in introductory physics often bring in a schema of mo-
tion that says objects tend to stop, I enthusiastically presented the result to Sagredo. He was
skeptical and argued “If you don’t tell them about friction, they wont know about it.” Sorry,
Sagredo. They may not know the word “friction” (they mostly do) or the rules of physics that
describe it (they mostly don’t), but they are very familiar with the fact that if you push a
heavy box across the floor, it stops almost as soon as you stop pushing it. They also know
that if they want to move, they have to exert an effort to walk and when they stop making
that effort, they stop.
“But,” responds Sagredo, “if you push a box very hard along a slippery floor it will keep
going for quite a distance. If you run and stop making an effort, you’ll continue going and
fall over. Surely they know those facts as well.” Absolutely right, Sagredo. But the problem is
that most students do not attempt to make a single coherent picture that describes all phe-
nomena. Most are satisfied with a fairly general set of often-inconsistent rules about “how
things work.”
Over the years there have been some disagreements among researchers as to the nature
of the schemas students bring with them to the study of introductory physics. Some re-
searchers suggested that students had “alternative theories”—reasonably self-consistent mod-
els of the world that were different from the scientist’s [McCloskey 1983]. But extensive work
by a number of different researchers (see in particular the work of McDermott [McDermott
1991] and diSessa [diSessa 1993]) suggests that student knowledge structures about physics
tend to be weak patterns of association that rarely have the character of a strong schema.15
Although I concur with this view, I note that occasionally a student’s schema may be
more coherent than we as scientists tend to give credit for, since we analyze the student’s views
through the filter of our own schemas. For example, a student who thinks about “motion”
and fails to separate velocity from acceleration may seem inconsistent to a physicist, whereas,
in fact, the student may feel he is consistent but may have a model that only works for a
rather limited set of specific situations and questions. As long as the student is either aware
14 Ilearned this from reading the paper of Halloun and Hestenes [Halloun 1985a] that also inspired Mazur—see
chapter 1.
15 By “strong” or “weak” we simply mean a high or low probability of activating a link to other related (and appro-
priate) items in a student’s schema.

of those limitations or not presented with situations in which that model doesn’t work, the
student can function satisfactorily.
There are two reasons why it is important for us to understand the knowledge and rea-
soning about the physical world that students bring with them into our classes. First, it helps
us understand what errors the students will commonly make and how they will misinterpret
what we say and what they read. We can then use this understanding to help us design both
new and improved instruction and better evaluations. I find understanding student thinking
particularly useful in helping me to answer students’ questions, both one-on-one during office
hours and during lecture. It is very easy to misinterpret a student’s question as being much
more sophisticated than it really is. Knowing common lines of reasoning or associations can
help probe the individual student as to what his or her problem really is. In lecture it can help
us understand what a question that seems a “stupid” question on its face may really be telling
us about the state of many of the students in our class and our interaction with them.
Second, the students bring into our class the basic resources from which they will build their
future knowledge. Since new knowledge is built only by extending and modifying existing
schema, students’ existing knowledge is the raw material we have to work with to help them
build a more correct and more scientific knowledge structure [Hammer 2000] [Elby 1999].
The knowledge and reasoning that students bring into our class have been analyzed in
three ways that are useful for us: (1) as common naïve conceptions, (2) in terms of primitive
reasoning elements, and (3) in terms of the way reasoning and knowledge are situated in
everyday real-life circumstances. The last-named is called situated cognition.
1. Robust reasoning structures: Common naïve conceptions

Occasionally, students’ patterns of associations concerning physical phenomena are strikingly
robust—they occur with a high enough probability in many contexts for us to refer to them
as mental models. In many cases, they contain inappropriate generalizations, conflations of
distinct concepts (such as treating velocity and acceleration as a single concept, “motion”), or
separations of situations that should be treated uniformly (such as treating a box sliding along
a rough floor and a rapidly moving baseball using different rules). When a particular mental
model or line of reasoning is robust and found in a significant fraction (say on the order of
20% of students or more), I refer to it as a common naïve conception. In the education re-
search literature, these patterns are often referred to as misconceptions, alternative conceptions,
or preconceptions, particularly when they lead to incorrect predictions or conclusions. I choose
the more descriptive term, “common naïve conceptions,” rather than the most common parl-
ance, “misconceptions,” both because it lacks the pejorative sting and because I want to em-
phasize the complexity of the student concept. Usually, these conceptions are not “just wrong.”
Students may be naïve, but they’re not fools. Their naïve conceptions are usually valuable and
effective in getting them through their daily lives. Indeed, most naïve conceptions have ker-
nels of truth that can help students build more scientific and productive concepts.
The presence of common naïve conceptions really isn’t so surprising if we think about
our students’ previous experience. Why should we be surprised that students think that any
moving object will eventually come to a stop? In their direct personal experience that is al-
ways the case. It’s even the case in the demonstrations we show in class to demonstrate the
opposite! When I slide a dry-ice levitated puck across the lecture table, I catch it and stop
it at the end of the table. If I didn’t, it would slide off the table, bounce, roll a short
distance, and stop. Every student knows that. Yet I ask them to focus on a small piece of the
demonstration—the stretch of about four or five seconds when the puck is sliding along
the table smoothly—and extend that observation in their minds to infinity. The student and
the teacher may focus on different aspects of the same physical phenomena.16
Many teachers show surprise when they learn the results of physics education research
demonstrating that students regularly generalize their naïve schemas incorrectly. Why should
it be surprising that students think cutting off part of a lens will result in only part of an im-
age being visible on the screen [Goldberg 1986]? Try looking through a magnifying glass!
(Yes, I know that’s not a real image.) Where do students get the idea that electricity is some-
thing that is used up in a resistor [Cohen 1983]? We’ve told them that you need circuits and
that currents flow in loops! Although we don’t always think about it, most of our students
have had extensive experience with electricity by the time they arrive in our classes. When I
said the current had to come in one wire and go out the other, one of my students com-
plained: “If all the electricity goes back into the wall, what are we paying for?”
Much of the effort in the published physics education research literature has been to doc-
ument the common naïve conceptions of introductory physics students and to develop in-
structional methods to deal with them. To get an introduction to this literature, consult the
papers listed in the Resource Letter given on the Resource CD [McDermott 1999]. A good
overview is given in the books [Arons 1990], [Viennot 2001], and [Knight 2002].
2. Modular reasoning structures: Primitives and facets

Perhaps the most extensive and detailed analysis of student reasoning in introductory physics
has been diSessa’s monumental work, “Toward an Epistemology of Physics” [diSessa 1993].17
In this study, diSessa analyzes the evolution of reasoning in mechanics of 20 MIT students
in introductory calculus-based physics. Although this is a fairly small number of students and
a rather narrow population, the care and depth of the analysis make it worthy of attention.18
Subsequent investigations show the presence of diSessa’s results in much broader populations.
As of this writing, diSessa’s approach has only rarely been applied to the development of new
curriculum. Nonetheless, because of the powerful insights it provides into student reasoning,
I believe it will be of use both in future curriculum reform and in research trying to under-
stand student thinking, and so I include a brief discussion of his ideas here.
DiSessa investigated people’s sense of physical mechanism, that is, their understanding of
“Why do things work the way they do?” What he found was that many students, even after
instruction in physics, often come up with simple statements that describe the way they think
things function in the real world. They often consider these statements to be “irreducible”—
as the obvious or ultimate answer; that is, they can’t give a “why” beyond it. “That’s just the
way things work,” is a typical response. DiSessa refers to such statements as phenomenologi-
cal primitives.
Some of these primitives may activate others with a reasonably high priority, but diSessa
claims that most students have rather simple schemas. Primitives tend to be linked directly
to a physical situation. They are recognized in a physical system rather than derived by a long
chain of reasoning.
16 This argument is made in a slightly different context in [KuhnT 1970].

17 Ashorter, more accessible introduction to diSessa’s ideas is given in [diSessa 1983].
18 Note that the point of this kind of study is to determine the range of possibilities, not their distribution among
a particular population. As a result, the small number of students in the study is not a serious drawback.
As an example, consider the response of a student to pushing a box on a rough surface.

The student might respond that “you need a big force to get it going” (overcoming : one in-
fluence may overpower another by increasing its magnitude), but then “you need a force to
keep it going” (continuous push: a constant effort is needed to maintain motion). DiSessa
identifies the parentheticals as primitives. Notice that the primitives are neither wrong nor
right in themselves. They are certainly correct in some circumstances, and diSessa points out
that experts use many primitives as quick and easy bits of compiled knowledge—but they are
linked so as to only be used in appropriate circumstances.
I like to add an additional structure. Some of diSessa’s phenomenological primitives are
very abstract (Ohm’s primitive, for example), and others refer to reasonably specific physical
situations (force as mover, for example). I prefer to distinguish abstract reasoning primitives
from those primitives applied in a particular context, which I refer to as facets (following
Minstrell).19 What I call an abstract reasoning primitive has a general logical structure, such as
“if two quantities x and y are positively related, more x implies more y.” What I call a facet
implies a mapping of the slots in the primitive into particular variables in a particular physi-
cal context. This is illustrated in Figure 2.5. As diSessa points out [diSessa 1993], there are a
very large number of facets, corresponding to the complexity of living in the world. In my for-
mulation, this complexity is seen as arising from mapping a reasonably small number of ab-
stract reasoning primitives (perhaps a few dozen) onto the large diversity of physical situations.
An example of mapping might be, “if the liquid is higher in one of the glasses, there is
more liquid in that glass.” In one of Piaget’s classic experiments, children are shown a vessel
containing some water. The water is then poured into a narrower vessel so that it rises to a
higher level. Before the age of about five years, most children say that the amount of water
has increased (because it’s higher) or decreased (because it’s narrower).20 Both those children
Primitives/Resources
Irreducible
functional pieces
based on direct
interpretation of
experience Context
Facets Both
Inferred physical internal
principles for and
specific situations external
Figure 2.5 A visual representation of the mapping of abstract primitives into specific facets in a par-
ticular physical context.
19 The term facet was introduced by Jim Minstrell [Minstrell 1992]. Minstrell listed large numbers of reasoning state-
ments, some correct, others not, that students gave as explanations for their observations or in predictions about
specific physical contexts.
20 This is not a simple failure to understand the meaning of the question. A child may get upset that a sibling is
“getting more” even if she is shown that when poured into a glass similar to their own they have the same heights.
who see it as more and those who see it as less are using what is essentially the same abstract
reasoning primitive, but with different mappings (focusing on height or width). After the age
of about five, children learn a “compensation” abstract reasoning primitive, something like
“if two effects act on a variable, one to increase it and the other to decrease it, those effects
tend to cancel.” Children reaching this ability to reason are said to have achieved Piagetian
conservation.
I’ve seen something very similar to this with engineers in first-year calculus-based physics
at the University of Maryland. We were discussing the collision of two objects of different
masses, and I asked about the relative size of the forces each object experiences. One group
of students said, “The larger objects feel a bigger force since they have bigger masses.” A sec-
ond group said, “The smaller objects feel a bigger force since they change their velocity more.”
Only a small number of students had reached the “Piagetian conservation” stage of activat-
ing the compensation implicit in Newton’s second law.
This kind of approach—analyzing the responses of our students in terms of primitives
and facets—helps us understand more clearly the kinds of reasoning we can expect. The crit-
ical realization that arises from this kind of analysis is that students’ common naïve concep-
tions are not simply wrong. They are based on correct observations but may be generalized
inappropriately or mapped onto incorrect variables. If we can extract elements that are cor-
rect from students’ common reasoning, we can build on these elements to help students re-
organize their existing knowledge into a more complete and correct structure.
3. Activating resources from everyday experience: Situated cognition

The primitives discussed above tend to refer to specific real-world situations as asked about
or observed in a physics class. A group of education specialists have focused on the difference
between day-to-day reasoning and the kind of reasoning taught and learned in schools.
Sagredo once stopped me in the hall after his introductory physics class for physics ma-
jors. “You’ll never guess what they couldn’t do today! I was talking about projectile motion
and asked them to describe what happens when a kicker kicks a football. I just wanted a de-
scription of the process—the ball goes up, travels a ways down the field, and comes down.
No one would say anything, even when I pressed them. Why couldn’t they give me a simple
description?” I suspect, Sagredo, that it was because they weren’t really sure what you wanted.
They might well have expected that you wanted some technical or mathematical description
in terms of forces, graphs, velocities, and accelerations. If they just said what you really
wanted—the simple day-to-day physical description of the process—they were afraid they
would look foolish. They may have been right to respond that way, given their previous ex-
perience with physics classes.
Most instruction in the United States today, despite reform efforts, continues to bear lit-
tle relation to students’ everyday lives. But many of the skills we are trying to teach can be
tied to reasoning skills that the students possess and use every day. An interesting example
comes from middle school arithmetic [Ma 1999]. Consider the following problem.
A group of students has 31⁄ 2 small pizzas, each whole divided into 4 parts. How many
students can have a piece?
The reasoning used by a student to solve this problem is quite a bit different from the algo-
rithm one learns for dividing fractions (31⁄ 2 / 1⁄ 4 4 31⁄ 2). The student might say something
like: “Each pie can serve 4 students, so the 3 pies can serve 12. The 1⁄ 2 pie can serve 2, so
a total of 14 can have a piece.” This reasoning, like the reasoning we use to solve the Coke/
Gin & Tonic problem in the previous section, relies on nonformal thinking that is linked to
our everyday social experience. Tying it to the dividing by fractions problem—showing that
division means finding how many times a part can be found in the whole—can help students
make sense of what division really means and why division is a useful concept.
The use of context knowledge to help solve problems is a common feature of how peo-
ple reason their way through situations in everyday life. A group of educators led by Jean
Lave and Lucy Suchman [Lave 1991] [Suchman 1987] places this cognitive fact at the cen-
ter of their educational reform efforts, creating cognitive apprenticeships and using situated cog-
nition. There is an extensive educational literature on this topic,21 and some dramatic im-
provements have been gained in children’s understanding of and effectiveness using arithmetic
by finding ways to use these everyday resources.
IMPLICATIONS OF THE COGNITIVE MODEL FOR INSTRUCTION:

FIVE FOOTHOLD PRINCIPLES
Any model of thinking is necessarily complex. We think about many things in many ways.
In order to find ways to see the relevance of these cognitive ideas and to apply them in the
context of physics teaching, I have selected five general principles that help us understand
what happens in the physics classroom.
1. The constructivism principle

4. The individuality principle
1. The constructivism principle

Principle 1: Individuals build their knowledge by making connections to existing knowl-
edge; they use this knowledge by productively creating a response to the information
they receive.
This principle summarizes the core of the fundamental ideas about the structure of long-term
memory and recall. The basic mechanism of the cognitive response is context-dependent as-
sociation. A number of interesting corollaries, elaborations, and implications that are relevant
for understanding physics teaching come from the constructivism principle.22
21 See,for example, [Lave 1991]. A very readable introduction to the subject is [Brown 1989].
22 These properties point out that our whole structure of patterns of association/schemas/mental models is a some-
what fuzzy one. The boundaries between the different structures are not sharply delineated. There are lots of exam-
ples of this sort of description in physics. Consider, for example, the excitations of a crystal lattice. We can describe
the excitations in terms of phonons or in terms of continuous waves. In some limiting cases, it is clear which de-
scription is the more useful; in others, they may overlap.
Implications of the Cognitive Model for Instruction: Five Foothold Principles • 31
Some of the characteristics of schemas clarify what is happening when students make
mistakes. Often in listening to my students explain what they think, I used to become con-
fused and sometimes irritated. How can they say x when they know the contradictory prin-
ciple y? Why can’t they get started on a problem when they certainly know the relevant prin-
ciple? They just told it to me two minutes ago! Why did they bring up that particular principle
now? It doesn’t apply here! The well-documented characteristics of mental structures described
in the first part of this chapter help us understand that these sorts of errors are natural and
to be expected.
It also makes us realize that we must get a lot more feedback than we traditionally get
if we want to understand what our students are really learning. Traditional testing often fails
to show us what students really think or know because many different schemas can produce
the correct solution to a problem. Even if a student goes through the same steps as we do,
there’s no guarantee that their schema for choosing the steps is the same as ours.23 I once
asked a student, who had done a homework problem correctly, to explain his solution. He
replied: “Well, we’ve used all of the other formulas at the end of the chapter except this one,
and the unknown starts with the same letter as is in that formula, so that must be the one
to use.”
Part of the way we fool ourselves with standard testing methods is that we are interested
“in what students know.” If they don’t access the right information in an exam, we give them
clues and hints in the wording to activate access. But since essential components of a schema
are the processes for access to information, we are not testing the students’ patterns of asso-
ciations if we narrow the context and provide detailed cues. The student “has” the informa-
tion, but it is inert and cannot be used or recalled except in very narrow, almost pre-
programmed situations.
To find out what our students really know, we have to find out what resources they are
bringing and what resources they are using. We have to give them the opportunity to explain
what they are thinking in words. We must also only give exam credit for reasoning and not
give partial credit when a student tries to hit the target with a blast of shotgun pellets and
accidentally has a correct and relevant equation among a mass of irrelevancies. To know
whether our students access the information in appropriate circumstances, we have to give
them more realistic problems—problems that relate directly to their real-world experience
and do not provide too many “physics clues” that specify an access path for them. (I’ll dis-
cuss the implications of this for assessment in chapter 4.)

The second principle reminds us of the nonuniqueness of the cognitive response and sets the
stage for the description of the dynamics of building mental structures.
Principle 2: What people construct depends on the context—including their mental

states.
It’s very easy to forget the warnings and drop back into the model that assumes students
either know something or don’t. Focusing on the context dependence of a response helps us
23 Thedifficulty is that the mapping from underlying schema to problem-solving steps is not one-to-one. A specific
example of this is given in [Bowden 1992].
keep in mind that the situation is not that simple. Nice examples of context dependence in
students’ responses to physics abound (although they are sometimes not presented that way).
One particularly clear example comes from the work of Steinberg and Sabella [Steinberg 1997]
[Sabella 1999]. At the end of the first semester of engineering (calculus-based) physics at the
University of Maryland, they gave a section of 28 students the (to an expert) equivalent pair
of questions, shown in Figure 2.6.
The first question is taken from the Force Concept Inventory (FCI) [Hestenes 1992].24
It is stated in common speech using familiar everyday objects. The second is a typical physics
class situation. It is abstract, and it involves idealized laboratory-style objects. The FCI was
given as an ungraded diagnostic at the end of classes25 and the problem as part of the final
exam one week later. Both test the students’ understanding of Newton’s first law.
Although 25 (90%) got the exam question correct, only 15 of the students (55%)
got the FCI question right. Nearly half of the students who succeeded with the problem in
the exam context missed the question presented in a nonphysics context (11/25 45%). In-
terviews with the students suggest that this is a real phenomenon, not just a result of an ad-
ditional week’s time to study.
An elevator (as illustrated) is being lifted up an elevator shaft by a steel cable.

When the elevator is moving up the shaft at a constant velocity (assume that any
frictional forces due to air resistance can be ignored):
a) the upward force on the elevator by the cable is greater than the Steel
downward force of gravity cable
b) the amount of upward force on the elevator by the cable is equal to the
downward force of gravity
c) the upward force on the elevator by the cable is less than the downward
force of gravity
d) it goes up because the cable is being shortened, not because of the
force being exerted by the elevator on the cable
e) the upward force by the elevator on the cable is greater than the downward
force due to the combined effects of air pressure and the force of gravity.
Ignore all friction and air resistance in this problem.
A steel ball resting on a small platform mounted to a hydraulic lift is being

lowered at a constant speed, as shown in the figure at right.
a) Draw a free-body diagram of the ball. Floor
Describe each type of force.
b) Compare the magnitudes of the forces you have drawn. Explain
your reasoning.
Figure 2.6 Two problems from [Steinberg 1997] demonstrating the context dependence of student responses.
24 The FCI is discussed in detail in chapter 5 and is included in the Resource CD associated with this volume.
25 These tests are often given as ungraded in order to encourage the students to give the answer they believe rather
than the answer they might think we want in physics class.
What we really want to help our students do is build their knowledge into a coherent
schema that is appropriately linked and that is triggered in a wide range of appropriate
contexts.

This principle deals with the dynamics of the mental state. It states that schemas are not only
the way that we organize our interactions with the world, but they also control how we in-
corporate new information and experiences [Bransford 1973].
Principle 3: It is reasonably easy to learn something that matches or extends an existing

schema, but changing a well-established schema substantially is difficult.
I pose a number of restatements and elaborations of this principle as corollaries to clar-

ify what it means for teaching.
Corollary 3.1: It’s hard to learn something we don’t almost already know.
All students have things they know (some of which may be wrong!), things they are a
bit familiar with, and things they have no knowledge about at all.
I like to look at this as an archery target. What they know is the bull’s-eye—a compact
black area; what they know a little about is a gray area surrounding the black; outside that is
a white “rest of the world” about which they are clueless. To teach them something, we do
best to hit in the gray. A class full of students is a challenge because all of their gray areas are
not the same. I want to hit as many of the grays as possible with each paint-tipped shaft of
information to turn gray areas black.26 This metaphor only describes some aspects of the pro-
cess. The real issue is that when we “hit in the gray,” the student has many appropriate links
that can be made to weave the new knowledge into their existing structure in an appropri-
ate way.
In communication studies, an important implication of this corollary is called the given-
new principle [Clark 1975] [RedishJ 1993]. It states that new information should always be
presented in a context that is familiar to the reader and that the context should be established
first. The analogous statement is very important in physics teaching, especially at the intro-
ductory level. As physicists with years of training and experience, we have a great deal of “con-
text” that our students don’t possess. Often we are as fish in water; unaware of having this
context and unaware that it is missing in our students.
We can cite a number of specifics that are violations of the given-new principle. One im-
portant example is that we often use terms that students are not familiar with—or that they
use in a different sense than we do. Lakoff and Johnson [Lakoff 1980], as a part of their study
of the way speakers of English build their meaning of the term force, classified the character-
istics of common metaphors using the term. Among their list of 11 characteristics, 8 involved
26 This picture also interacts strongly with the social learning principle discussed below. Items in the gray are
those that the student can learn via social interactions with teachers or more expert students. The followers of the
Russian psychologist Lev Vygotsky give the gray region the unfortunate name “zone of proximal development”
[Vygotsky 1978].
the will or intent of an individual! But most of us are so familiar with the technical meaning
of “force” that we are surprised to learn that a significant fraction of our introductory students
do not believe that a table exerts a force on a book it is supporting [Minstrell 1982]. Why
doesn’t the book fall through? The table is just “in the way.” (This issue is discussed in more
detail under the heading “Bridging” later in this chapter.)
The problem caused by the interpretation of common speech words for technical ones
is not simple. I know that the terms “heat” and “temperature” are not really distinguished in
common speech and are used interchangeably for the technical terms “temperature” (average
energy per degree of freedom), “internal energy”, and “heat flow” (flow of internal energy
from one object to another). In one class, I stated this problem up front and warned my stu-
dents that I would use the terms technically in the lecture. Part way through I stopped, re-
alizing that I had used the word “heat” twice in a sentence—once in the technical sense, once
in the common speech sense.27 It’s like using the same symbol to stand for two different
meanings in a single equation. You can occasionally get away with it,28 but it really isn’t a
good idea!
Putting new material in context is only part of the story. Our students also have to see
the new material as having a plausible structure in terms of structures they know. We can
state this as another useful corollary.
Corollary 3.2: Much of our learning is done by analogy.
This and the previous corollary make what students know at each stage critical for what
we can teach them. Students, like everyone else, always construct their knowledge, and what
they construct depends on how what we give them interacts with what they already have. This
has important implications for the development of instructional techniques that help students
overcome strong misconceptions. (See the discussion of “bridging” in a later section.)
One implication of these results is that we should focus on building structures that are
useful for our students’ future learning. I state this as a third corollary.
Corollary 3.3: “Touchstone” problems and examples are very important.
By a touchstone problem,29 I mean one that the student will come back to over and over
again in later training. Touchstone problems often become the analogs on which they will
build the more sophisticated elements of their schemas. It becomes extremely important for
students to develop a collection of a few critical things that they really understand well.30
These become the “queen bees” for new swarms of understanding to be built around. I
27 “If there is no heat flow permitted to the object, we can still heat it up by doing work on it.” ⁄

eh
28 I have seen colleagues write the energy levels of hydrogen in a magnetic field as Enlm En mB where the
2m
m in the denominator is the electron mass and the one in the numerator is the z-component of the angular mo-
mentum. Most physicists can correctly interpret this abomination without difficulty.
29 In his discussion of scientific paradigms, T. S. Kuhn refers to these problems as exemplars [KuhnT 1970].
30 In addition to giving them centers on which to build future learning, knowing a few things well gives the student
a model of what it means to understand something in physics. This valuable point that has been frequently stressed
by Arnold Arons [Arons 1990]. It is an essential element in developing scientific schemas.
believe the sense that some superficially uninteresting problems serve this role is the reason
they have immense persistence in the community. Inclined plane problems really aren’t very
interesting, yet the occasional suggestions that they be done away with are always resisted
vigorously. I think the resisters are expressing the (often unarticulated) feeling that these are
the critical touchstone problems for building students’ understanding of vector analysis in
the plane.
Corollary 3.3 is one reason we spend so much time studying the mass on a spring. Springs
are of some limited interest in themselves, but small-amplitude vibrations are of great gen-
eral importance. The spring serves as a touchstone problem for all kinds of harmonic oscil-
lations from electrical circuits up to quantum field theory.
Analyzing a curriculum from the point of view of the schema we want students to de-
velop, their preexisting schemas, and touchstone problems that will help them in the future
can help us understand what is critical in the curriculum, which proposed modifications could
be severely detrimental, and which might be of great benefit.
Combining these ideas with the idea of associations discussed under Principle 1 leads us
to focus on the presence of a framework or structure within a course. It suggests that build-
ing a course around a linked series of touchstone problems could be of considerable assis-
tance in helping students understand the importance and relevance of each element. Such a
structure is sometimes referred to as a story line.
Unfortunately, if students are not blank slates, sometimes what is written is—if not en-
tirely wrong—inappropriate for future learning in physics.31 Then it can seem as if we have
run into a brick wall. This brings us to the next corollary.
Corollary: 3.4: It is very difficult to change an established mental model.
Traditionally, we’ve relied on an oversimplified view of Principle 1, the constructivism

principle, to say: “Just let students do enough problems and they’ll get the idea eventually.”
Unfortunately, this simple translation of the principle doesn’t necessarily work. Although prac-
tice is certainly necessary to help students compile skills into easily retrievable knowledge,
there is no guarantee that they will link those skills into a structure that helps them to un-
derstand what’s going on and how to use the basic concepts appropriately.
The limitations of doing lots of problems were investigated in a study done in Korea.
Eunsook Kim and Jae Park looked at the response of 27 students in an introductory college
physics class to the FCI [Kim 2002]. American students at large, moderately selective state
universities (such as the University of Maryland, the Ohio State University, or the University
of Minnesota) who have taken one year of high school physics score an average of 45–50%
on this test before beginning a calculus-based physics class. The students Kim studied had
taken an apparently much more rigorous high school physics program in which each student
had done an average of about 1500 problems (ranging between 300 and 2900) end-of-
chapter problems. In a typical American high school class, students will do 300 to 400 such
problems. Despite doing 5 to 10 times as many problems as American students, the students
31 Perhaps a palimpsest is a better metaphor for a student’s mind than a blank slate. According to the American Her-
itage Dictionary, a palimpsest is “a manuscript, typically of papyrus or parchment, that has been written on more
than once, with the earlier writing incompletely erased and often legible.”
still had substantial conceptual difficulties with fundamental concepts of mechanics at rates
comparable to those seen with American students. There was little correlation between the
number of problems the students had done and the conceptual understanding displayed.
This study and others like it are a warning against relying on the idea that “repetition
implies effective learning”—that is, that frequent repetition of a particular type of activity is
all that is needed to produce a useful functional schema. Repetition is necessary to create
compiled knowledge, but it is not sufficient. For effective usage, the compiled element needs
to be linked into a coherent schema about the subject.
Once students learn how to do problems of a particular type, many will learn nothing
more from doing more of them: New problems are done automatically without thinking.
This also means that testing by varying homework problems slightly may be inadequate to
probe the student’s schemas. More challenging tests involving a variety of modes of thinking
(problem solving, writing, interpreting, organizing) are required. Such testing is discussed in
detail in chapter 4.
It has been demonstrated over and over again that simply telling students some physics
principle doesn’t easily change their deep ideas.32 Rather, what often happens is that instead
of changing their schema substantially, a poorly linked element is added with a rule for us-
ing it only in physics problems or for tests in one particular class. This and the fact that a
schema can contain contradictory elements is one possible reason “giving more problems” is
often ineffective.
A few years ago, I heard a lovely anecdote illustrating the barriers one encounters in try-
ing to change a well-established mental model.33 A college physics teacher asked a class of
beginning students whether heavy objects fall faster than light ones or whether they fall at
the same rate. One student waved her hand saying, “I know, I know.” When called on to ex-
plain she said: “Heavy objects fall faster than light ones. We know this because Galileo dropped
a penny and a feather from the top of the leaning tower of Pisa and the penny hit first.” This
is a touchstone example for me. It shows clearly that the student had been told—and had
listened to—both the Galileo and the Newton stories. But she had transformed them both
to agree with her existing mental model.34
Fortunately, mechanisms are available for helping students restructure even well-
established mental models; these methods are discussed later in this chapter.
4. The individuality principle

One might be tempted to say: Fine. Let’s figure out what the students know and provide
them with a learning environment—lectures, demonstrations, labs, and problems—that takes
them from where they are to where we want them to be. Since we all know that a few
32 See the papers referred to in the annotated bibliography [McDermott 1999] and in the review papers [McDermott
1991], [Reif 1994], and [Redish 1999].

33 Audrey Champagne, private communication.
34 We should not lose sight of the fact that the student’s mental model in this case is in fact correct. We observe that
lighter objects do fall more slowly than heavy ones if they fall in air, and few of us have much direct experience with
objects falling in a vacuum. But for reasonably dense objects falling for only a few seconds, the difference is small,
so that this observation does not yield a useful idealization. The observation that objects of very different mass fall
in very nearly the same way does.
students get there from here using our current procedures, why can’t we make it work for
all of them? We do in fact know now that the right environment can produce substantially
better physics learning in most of the students taking introductory university physics.35
But my fourth principle is a word of warning that suggests we should not be looking for a
“magic bullet.”
Principle 4: Since each individual constructs his or her own mental structures, different
students have different mental responses and different approaches to learning. Any
population of students will show a significant variation in a large number of cognitive
variables.
I like to call this principle the individuality or distribution function principle. This re-
minds us that many variables in human behavior have a large natural line width. The large
standard deviation obtained in many educational experiments is not experimental error; it is
part of the measured result! As physicists, we should be accustomed to such data. We just
aren’t used to its being so broad and having so many variables. An “average” approach will
miss everyone because no student is average in all ways.36
In addition to the fact that students have different experiences and have drawn different
conclusions from them, their methods of approach may differ significantly. I state this as a
corollary.
Corollary 4.1: People have different styles of learning.
There is by now a vast literature on how people approach learning differently. Many
variables have been identified on which distributions have been measured. These include
authoritarian/independent, abstract/concrete, and algebraic/geometric, to name a few. The
first variable means that some students want to be told, while others want to figure things
out for themselves. The second means that some students like to work from the general to
the specific, and some the other way round. The third means that some students prefer to
manipulate algebraic expressions, while others prefer to see pictures. Many of us who have
introduced the computer in physics teaching have noted that some students want to be guided
step by step; others explore everything on their own. These are only a few of the variables.
For some good analyses of individual cognitive styles and differences, see [Gardner 1999],
[Kolb 1984], and [Entwistle 1981].
Once we begin to observe these differences in our students, we have to be exceedingly
careful about how we use them. A preference does not mean a total lack of capability. Stu-
dents who prefer examples with concrete numbers to abstract mathematical expressions may
be responding to a lack of familiarity with algebra rather than a lack of innate ability. Many
of our students’ preferences come from years of being rewarded for some activities (such as
35 Asexamples, see the references cited in the section “Research-Based Instructional Materials” in [McDermott 1999].
This annotated bibliography is also included on the Resource CD.
36 This is analogous to the story of the three statisticians who went hunting deer with bows and arrows. They came
across a large stag and the first statistician shot—and missed to the left. The second statistician shot and missed to
the right. The third statistician jumped up and down shouting “We got him!”
being good memorizers) and chastised for others (such as asking questions the teacher couldn’t
answer). Expanding our students’ horizons and teaching them how to think sometimes re-
quires us to overcome years of negative training and what they themselves have come to be-
lieve are their own preferences and limitations.
An important implication is the following:
Corollary 4.2: There is no unique answer to the question: What is the best way to teach
a particular subject?
Different students will respond positively to different approaches. If we want to adopt

the view that we want to teach all our students (or at least as many as possible), then we must
use a mix of approaches and be prepared that some of them will not work for some students.
We need to answer the question: What is the distribution function of learning characteristics
that our students have in particular classes? Although some interesting studies have been done
over the years, the implication for instruction in physics is not well understood.37
Another implication that is very difficult to keep in mind is:
Corollary 4.3: Our own personal experiences may be a very poor guide for telling us the
best way to teach our students.
Physics teachers are an atypical group. We “opted in” at an early stage in our careers be-
cause we liked physics for one reason or another. We then trained for up to a dozen years be-
fore we started teaching our own classes. This training stretches us even farther from the style
of approach of the “typical” student. Is it any wonder that we don’t understand most of our
beginning students and they don’t understand us?
I vividly recall a day a few years ago when a student in my algebra-based introductory
physics class came in to ask about some motion problems. I said: “All right, let’s get down to
absolute basics. Let’s draw a graph.” The student’s face fell, and I realized suddenly that a
graph was not going to help him at all. I also realized that it was going to be hard for me to
think without a graph and to understand what was going through the student’s mind. I never
minded doing without derivatives—motion after all is the study of experimental calculus, and
you have to explain the concept (maybe without using the word “derivative”) even in a non-
calculus-based class. But I can’t remember a time when I couldn’t read a graph, and I have
found it difficult to empathize with students who come to physics and can’t read a graph or
reason proportionately. It takes a special effort for me to figure out the right approach.
This is very natural given the earlier principles. Our own schemas for how to learn come
from our personal reactions to our own experiences. However, to reach more of our students
than the ones who resemble ourselves, we will have to do our best to get beyond this mind-
set. It makes the following principle essential.
Corollary 4.4: The information about the state of our students’ knowledge is contained
within them. If we want to know what they know, we not only have to ask, we have to
listen!
37 See [Kolb 1984].
One point I want to stress about the individuality principle is in the idea expressed by
its final sentence: Any population of students will have a significant variation in a large num-
ber of cognitive variables. We have a tendency, especially in physics, to classify students along
a single axis that we think of as “intelligence.” I’ve heard Sagredo say, “Well, most of my stu-
dents are having trouble, but the smart ones get it.” In cognitive science, there have been vig-
orous arguments for a number of years now as to whether there is a single variable (referred
to as “g”) that describes intelligence, or whether what we call intelligence consists of a num-
ber of independent factors.38 The literature on this subject is quite complex, and I do not
pretend to be able to evaluate it. However, whether or not intelligence is a unary concept,
success in physics—or in any scientific career—relies on much more than intelligence. I have
followed with interest the careers of many of my physics classmates from college and gradu-
ate school for many decades, and one thing is absolutely clear. The students who were the
“brightest” in doing their schoolwork were not necessarily the ones who went on to make the
most important contributions to physics. Creativity, persistence, interpersonal skills, and many
other factors also played large roles.39 This point is discussed again later in the next chapter
on the hidden curriculum.

With the fifth principle, I go beyond single individuals and consider their relations to oth-
ers as a part of their learning. This principle is based on the work on group learning that
builds on the ideas of the Russian psychologist, Lev Vygotsky. These ideas have had a pro-
found impact on modern theories of teaching and learning [Vygotsky 1978] [Johnson 1993].
Principle 5: For most individuals, learning is most effectively carried out via social
interactions.
The social learning principle is particularly important for physicists to keep in mind.
Physicists as a group are highly unusual in many ways. In my experience, they tend to be in
the extreme tails of distributions of curiosity, independence, and mathematical skills. They
also tend to be highly self-sufficient learners. I once heard David Halliday remark that what
he enjoyed most as a student was sitting down by himself alone in a quiet room with a physics
text and going “one-on-one” with the authors of the book—trying to understand them and
figure out what they were saying. Many of us have similar inclinations. Physicists as a group
seem to be selected for the character of being able to learn on their own. But in examining
my experiences of this type, I have decided that my “learning on my own” involves an abil-
ity to create an “internalized other”—to take a variety of viewpoints and to argue with my-
self. This is not a commonly found characteristic and should not be assumed in a general
population of students.
38 Fora discussion and for references to this literature, see [Gardner 1999].
39 The range of important variables provides a basis for what we might call Garrison Keillor’s Corollary. “All students
are above average—on some measure.”
SOME GENERAL INSTRUCTIONAL METHODS

DERIVED FROM THE COGNITIVE MODEL
Many instructional methods have been developed based on the cognitive models discussed
above. Two that will be relevant for us later in our discussions of specific curricula are cog-
nitive conflict and bridging. The cognitive conflict method is used when an inappropriate gen-
eralization or incorrect association has become particularly robust and difficult to change.
The bridging method relies on the explicit idea that the students bring useful knowledge re-
sources to their learning of physics and attempts to explicitly activate those resources in ap-
propriate ways.
Cognitive Conflict
Common naïve conceptions can be strikingly robust. At the beginning of my unit on direct
current circuits in my calculus-based engineering physics class, I gave the students the prob-
lem shown in Figure 2.7.
This problem is by now rather traditional. It appears in sixth grade science texts and in
many high school physics texts. My discussion of circuits occurred at the beginning of our
third semester of engineering physics. Most of the students were second-semester sophomores,
and many of them were electrical engineering students. More than 95% of them had taken
(and done well in) high school physics. Yet only about 10-15% of them were able to solve
this problem before instruction.
Sagredo complained that this was “a trick question.” He said, “I’ll bet many graduate
students in physics would miss it. You have to find a clever way to make it light—touching
the bulb’s tip to one end of the battery to make a contact without a wire. Then you can use
the wire to close the circuit.”
Sagredo’s correct answer is shown at the left of Figure 2.8. He is right that many physics
grad students will miss the problem on the first try (and many professors as well), but the spe-
cific wrong answers given by students and by experts show that something different is going
on. The experts who get it wrong give the answer, “No. It can’t be done. You need a closed
circuit and that takes two wires.” My engineering students’ wrong answers were, “Sure you
Duracell
Given the three objects shown above, a battery, a bulb, and a single piece of wire, can
you make the bulb light without cutting the wire? If you can, show how. If you can’t,
explain why not.
Figure 2.7 A problem introductory students often have difficulty with.

Some General Instructional Methods Derived From the Cognitive Model • 41
Duracell
Duracell
Duracell
Figure 2.8 Correct answer (on left) to battery-bulb-wire problem and the two most common incor-
rect student answers.
can. It’s easy.” One-third of the students gave the answer shown in the middle, and one-third
gave the answer shown at the right. About half of the rest gave the correct answer. The rest
either left it blank or gave uninterpretable answers. Very few students gave the “expert’s error.”
Many students showed the common naïve conception of electricity as a source of energy
that can be “drawn” directly from an electric power source or “tapped into” by touching. Stu-
dents’ naïve conceptions about electric currents are documented in the research of the Uni-
versity of Washington Physics Education Group (UWPEG) [McDermott 1992]. The group
also reports a lesson they have developed to address this issue [Shaffer 1992]. The lesson is
delivered in a “Tutorial”—a structure that has replaced recitations at the University of Wash-
ington. These and related materials are a part of the Physics Suite and are discussed in detail
in chapter 8.
The model frequently used in Tutorials is cognitive conflict in the form elicit/confront/
resolve/reflect. In the first Tutorial on direct currents, the lesson begins with the question shown
above (given on a pretutorial ungraded quiz40 during lecture). When the students get to the
Tutorial, each group gets a battery, a bulb, and a single wire.
When I gave this lesson in my classes, about half of the students expected to be able to
light the bulb using one of the two arrangements at the right of Figure 2.8. I particularly re-
member one student who came up to me complaining that the equipment she had been given
had to be defective. (She was certain that the middle arrangement shown in Figure 2.8 should
work. “After all,” she said, “that’s the way a flashlight works, isn’t it?”) She insisted on hav-
ing a new bulb and a fresh battery.
The subsequent discussions with their colleagues and with the facilitators (and the rest
of the lesson, which elaborates on the point and reconsiders it in a variety of contexts) help
students to resolve the conflict between their model and their observations, and to reflect
40 One reason for not grading the quiz is to encourage students to look for what they actually think is plausible rather
than to try to guess what the teacher wants. The solutions also are not posted since the point is to get students think-
ing about the reasoning (during the Tutorial period) rather than focusing on the answer.
upon the implications. After Tutorial instruction, students did significantly better (75%
success) on other (more complex) questions probing similar issues than students with tradi-
tional instruction ( 50% success). Many other examples exist of successful learning pro-
duced through lessons based on cognitive conflict.
Bridging
The use of cognitive conflict as the primary instructional tool in the classroom can lead to
difficulties because it is a rather negative approach. A colleague who had instituted a reform
procedure in a high school physics class that relied heavily on the cognitive-conflict method
reported to me that one of his students responded to a quiz: “Oh great. Here’s another test
to show us how stupid we are about physics.” Once students are convinced they cannot do
physics, it is extremely difficult to teach them anything more than rote memorization.41
However, if we consider the student’s starting knowledge as a resource and begin with what
is right about student reasoning, we can make them feel considerably better about themselves
and about their ability to do physics. To see how this might work, let’s consider an example.
John Clement has proposed building on the correct aspects of students’ naïve concep-
tions by finding a series of bridges or interpolating steps that help students transform their
mental models to match the accepted scientific one [Clement 1989]. The starting point for
a bridge should be something that the students know well that is correct—which Clement
refers to as an anchor. An example of bridging is in Clement’s building of a lesson to respond
to the common naïve conception about normal forces.
In his classic paper, “Explaining the ‘At Rest’ Condition of an Object,” Jim Minstrell
documented that more than half of the students in his high school physics class did not be-
lieve that an inanimate supporting object could exert a force [Minstrell 1982]. They inter-
preted what a table does to a book resting upon it as “simply preventing it from falling.” They
showed that the pattern of association triggered by the word “force” had a strong link to the
idea of will or intent and hence, to something produced by an “active” object, most frequently
an animate one. Students seem to be bringing up a “blocking” reasoning primitive rather than
a “springiness” one. This result has subsequently been confirmed to occur in a significant frac-
tion of introductory physics students.
Clement looked for examples in which students had correct intuitions (primitive re-
sponses or “gut feelings”) that could be used to build on. He suggests that a useful starting
point for a bridging lesson should have the following characteristics.
1. It should trigger with high probability a response that is in rough agreement with the
results of correct physical theory.
2. It should be concrete rather than abstract.
3. The student should be internally confident of the result; that is, students should
strongly believe in the reasoning by themselves and not by virtue of authority.
Clement came up with two possible anchors for the situation described by Minstrell.
• Holding up a heavy book in your hand

• An object being held up by a reasonably soft spring
41 These issues are discussed in more detail in chapter 3.
TABLE 2.1 Results of a Bridging Lesson

Fraction of possible
Pre-test average Post-test average gain achieved
Control group (N 55) (17 1)% (45 2)% 0.34
Experimental group (N 155) (25 2)% (79 1)% 0.72
To Clement’s surprise, the spring served as a much better starting point than the book. This
stresses the fact that we cannot assume that an example that seems an obvious anchor to us
will work for our particular population. They must be tested. Our individuality principle also
reminds us not to expect that a particular example will serve for all students. We may have
to use a number of different anchors to reach most students.
Clement tested this approach in four high schools in Massachusetts [Clement 1998].
There were 155 students in the experimental group and 55 in the control. The same pre-
and post-tests were given to all students. On the issue of static normal forces, the gains of
the groups were as shown in Table 2.1. The experimental group was taught with a bridging
lesson; the control class with a traditional one. Evaluation was done with six questions that
were part of a longer test. (Errors are std. error of the mean.) Similar results were reported
for bridging strategies in the areas of friction and dynamic use of Newton’s third law in
collisions.42
Restricting the frame

The real world, with all its complexity, is too much for us to handle all at once. Long-term
and working memories function together to parse the visible world into the pieces relevant
for functioning in any particular situation. Many students, when facing physics problems,
have considerable difficulty figuring out what is important and what is not. Helping them
build appropriate templates and associations is an important part of what physics instruction
is trying to accomplish.
Limiting our view of what we want to analyze is an essential part of the scientific en-
terprise. One of Galileo’s greatest contributions to science was his ability to step back from
the Aristotelian attempt to produce a grand synthesis for all motion and to focus on under-
standing well how a few simple phenomena really worked—an object on an inclined plane,
the pendulum, a falling body. The synthesis comes later, once you have some solid bricks to
build into a more coherent structure. When Newton synthesized a theory of motion 50 years
later, he created a limited synthesis—a theory of motion, but not a single theory that suc-
cessfully described, for example, light, heat, and the properties of matter at the same time.
The scientific structure grows at different paces and in different places, with pieces being con-
tinually matched and modified to create more coherent and useful maps.
Similarly, in teaching introductory physics, we also have to restrict our considerations to
a piece of the entire picture. Our goal is not to present the most coherent picture of the en-
tire physical knowledge system that we as professionals have been able to construct, even
42 For a discussion of the lessons and the controls, see [Clement 1998].
though this might be an intellectually enticing and enjoyable goal. The mental structure that
we’ve created for ourselves to describe physics has been built up over many years—perhaps
decades—and has resulted from continual transformations that have been made not only to
the physics, but to ourselves.
Even in solving a single problem in introductory physics, we must go through the pro-
cess of restricting consideration to a piece of the world and of limiting those aspects we want
to consider. I refer to selecting a limited piece of the world to view through a frame as cat tele-
vision. My cat (see Figure 2.9) very much enjoys viewing a small piece of the world through
a window (as have many other cats I have known) and can become quite addicted to it. When
we drove across the country for a sabbatical visit with the cat in the car, in every motel room
he insisted on finding a place where he could look out the window and see where in the world
we were. The real TV doesn’t interest him at all. (The same is true for many physicists.)
Once we have drawn our frame—chosen our channel on cat TV—we still have work to
do before we can start drawing a representation. We have to choose what to pay attention to
and what to ignore. I call this process creating the cartoon. I mean “cartoon” here in the sense
of an artist’s sketch, drawn in preparation to creating a painting or mural. The process of de-
ciding what to keep and what to ignore is a difficult one, and one that is often glossed over
by physicists who already know what to consider. (An example is Bill Amend’s FoxTrot
cartoon—in the other sense—shown in Figure 2.10.)
Changing what one decides to look at in a real-world physical phenomenon can be as-
sociated with a major advance in science. In his Structure of Scientific Revolutions, T. S. Kuhn
[KuhnT 1970] describes the paradigm shift associated with Galileo’s observation of the pen-
dulum. The story goes that Galileo was sitting in church, bored with an extended sermon,
perhaps, when a gust of wind started a chandelier swinging. An observation that could be
made on the chandelier is that it eventually comes to rest. One could infer that this supports
the Aristotelian position that objects naturally come to rest (and seek the lowest point). But
Galileo timed the swing of the chandelier using his pulse and noted that the period of the
Figure 2.9 “Cat television.” Cats often enjoy observing a

small, framed “piece of the world.” Similarly, a scientist must
narrow his or her focus in a particular set of issues in order
to make progress. I call this “choosing a channel on cat TV.”
Figure 2.10 Assumptions in traditional physics problems are often tacit but important.
swing did not change as the amplitude got smaller. He then realized that this could be in-
terpreted as saying that the natural state of the object was to continue to oscillate. In the Aris-
totelian view, the coming to rest is taken as fundamental, and the oscillation is seen as a de-
viation. In the Galilean view, the oscillation is seen as fundamental, and the damping is seen
as peripheral. There is no a priori way to decide which view is better. One must have more
information as to how these assumptions play out and how productive they are in generat-
ing deeper understandings of a wider class of phenomena.
When students view a subject or a problem in physics, they are often unclear as to how
to link what the physicist sees in the problem with their personal experience. A real-world
problem involves friction, massive pulleys, cartwheels that have significant moments of iner-
tia, etc., etc., etc. Knowing what to ignore is an important part of learning to think about sci-
ence, and it should not be treated as trivial. Sagredo often teaches his students “the first step
in doing a physics problem is to draw a diagram,” but expresses frustration that often the
students “can’t even do that.” I have seen students who, when given that instruction, draw a
3-D diagram when a 2-D one will do, or who carefully draw a realistic car, person, or horse
when the problem could be done by using a block. Is it obvious, when asked to “consider a
car rolling down a hill,” that it doesn’t matter whether the car is a Porsche or a VW bug or
a block of wood with wheels? I conjecture that our ignoring this step is one reason that stu-
dents don’t always see the relation of physics to their real-world experience (see chapter 3).
Multiple representations
In science we use a dizzying variety of representations43 that may have a variety of positive
cognitive effects. First, the use of multiple representations (visual and verbal, for example)
may help us make better use of our working memory. Second, different representations as-
sociate more naturally with different features of the data/situation we are trying to describe.
As a result, the use of multiple representations can be effective in building links between dif-
ferent aspects of the situation. Some of the representations we use in physics include
• words
• equations
• tables of numbers
• graphs
• specialized diagrams
For the expert, each representation expresses some characteristic of the real-world system
more effectively than others, and the translation and links between them help build up a
strong and coherent mental model. But each representation takes some time and effort to
learn. If students haven’t yet learned to “read” a representation, that is, if their knowledge of
that representation hasn’t been compiled into an easily processed chunk, the translation of
the representation can take up too much working memory to allow the students to make ef-
fective use of the representation. Some may have mastered one or more of these representa-
tions through experiences they had before coming to our class. Some may have a learning
style that favors a verbal or a visual representation. Some may think they have trouble with
one or more of our representations and actively avoid thinking about them [Lemke 1990].
I represent our process of setting up a problem or scientific exploration with the diagram
shown in Figure 2.11. We begin by picking a channel of cat television with a specific real-
world situation considered in a specific physics context—for example, the mechanics of a car
rolling down a hill. We then make a selection of what we want to look at first—a simplified
image that corresponds to a box sliding down a frictionless inclined plane—what I called
above: creating a cartoon. Is this a good model for an SUV rolling down a rocky mountain
trail? Perhaps not. But it might be a good place to start for a car with smaller wheels rolling
down a paved road.
Once we have our cartoon, we then express our knowledge of the situation in a variety
of ways, using linked multiple representations. These present the information in different
ways, enabling different ways of thinking about and linking the information.
There is considerable evidence in the research literature documenting student difficul-
ties with representation translation [Thornton 1990] [Beichner 1994]. Students often see, for
example, the drawing of a graph as the solution to a problem—something the teacher asked
them to do—instead of a tool to help them understand and solve a more complex problem.
43 See, for example, the detailed analysis of a chemistry and physics class in [Lemke 2000].
Rethinking the Goals of Physics Instruction • 47
As the car rolls

Words down the hill it
picks up speed
Equations v = g sin θt
t x v
Numbers 0 0 0
0.1 0.05 0.1
Graphs
Diagrams
Figure 2.11 Physicists use a large number of different representations to describe events and processes
in the real world. Learning to handle these representations and to translate them is an important—and
difficult—step for introductory physics students.
In order to use multiple representations in our physics classes, it is important that we be

aware of their difficulty for students and that we help those students both to learn the physics
and to learn to make the connection to the modes of thinking they are less comfortable with.
RETHINKING THE GOALS OF PHYSICS INSTRUCTION

Putting our students’ learning into a cognitive framework helps us begin a more detailed and
constructive dialog in the community of physics teachers on what we want to accomplish
with our physics instruction. There is more to learning physics than placing check marks on
the table of contents of a text. Many of the most important results of our instruction are not
associated with particular physics content. Even those goals that are associated with particu-
lar content can now be seen in a different way, given our understanding of student thinking.
In this section, I try to explicate some of the “hidden” goals of attitudes and skills that we
might like our students to attain in an introductory physics course—goals that are rarely dis-
cussed and whose attainment is often both strongly desired and taken for granted.
“Wait!” Sagredo interjects. “This talk of attitudes and skills is all very well, and I hope
my students will develop them. But the physics content is important and we shouldn’t lose
sight of it.” An excellent point, Sagredo. Let’s start with what our learning theory says about
learning physics content itself. Then we can consider attitude goals and skill development.
Extended content goals

Learning physics content means much more than memorizing a lot of independent defini-
tions and equations. We want our students to understand enough about what the physics
means to be able to understand what problems are about and what their answers imply; we
want them to understand how the physics they are learning fits—both with other physics
they have learned and with their personal experience with the world; and we want them to
be able to use their knowledge effectively in solving problems. I refer to these three goals as
concepts, coherence, and functionality.
Goal 1: Concepts—Our students should understand what the physics they are learning
is about in terms of a strong base in concepts firmly rooted in the physical world.
From our cognitive model we know that students can attach new knowledge to their ex-
isting knowledge structures as something separate and only relevant for the context of solv-
ing problems in a physics class. We want students to not only compile the mathematical ma-
nipulations associated with solving physics problems (i.e., learn to use them without thinking
about them); we want them to understand what the physics is about.
To achieve this goal, students have to make sense of physics concepts—the ideas and def-
initions that map abstract physics descriptions onto real things in the physical world. In or-
der to help them reach this goal through instruction, it often helps to motivate the need for
a concept before introducing the definition through direct observation of phenomena. Arnold
Arons called this “idea first, name afterwards” and stressed the value of operational defini-
tions in physics instruction. He has a nice description of this in his book [Arons 1990].
In my own courses, I indicate from the first day that we will operate under the precept “idea
first and name afterwards,” and that scientific terms acquire meaning only through the descrip-
tion of shared experience in words of prior definition. When students try to exhibit erudition
(or take refuge from questioning) by name dropping technical terms that have not yet been de-
fined, I and my staff go completely blank and uncomprehending. Students catch on to this game
quite quickly. They cease name dropping and begin to recognize, on their own, when they do
not understand the meaning of a term. Then they start drifting in to tell me of instances in
which they got into trouble in a psychology or sociology, or economic, or political science course
by asking for the operational meaning of technical terms. [Arons 1990]
While presenting “idea first, name second” is a good start, it’s rarely sufficient to get students
to develop a good understanding of physics concepts. A considerable amount of the effort in
physics education research over the past decade has been devoted to the development of ef-
fective instructional techniques for helping students build their conceptual understanding.
For more details on these specific content issues, see [Arons 1990] and the references in the
Resource Letter [McDermott 1999] given on the resource CD.
Goal 2: Coherence—Our students should link the knowledge they acquire in their physics
class into coherent physical models.
A major strength of the scientific worldview is its ability to describe many complex phe-
nomena with a few simple laws and principles. Students who emphasize science as a collec-
tion of facts may fail to see the integrity of the structure, an integrity that is both convinc-
ing and useful in solving problems. The lack of a coherent view can cause students many
difficulties, including a failure to notice errors in their reasoning and an inability to evaluate
a recalled item through cross-checks.
Let’s recall Principles 1 and 2: Students will put what we give them into their knowl-
edge structure and integrate it into their existing knowledge structure in some way of their
Rethinking the Goals of Physics Instruction • 49
own. Whatever it is that they know of any particular content, they may or may not create
coherent schemas with appropriate connections that are activated with a high probability in
appropriate contexts. We want our students to not simply “get the content” but to build their
understanding of that content into an accurate and effective mental model.
Goal 3: Functionality—Our students should learn both how to use the physics they are
learning and when to use it.
For most of our populations of introductory physics students, we don’t just want them
to know the physics content but to be able to do something with it. My second cognitive prin-
ciple suggests that in addition to having students master the physics content and see that it
makes sense, we also want students’ knowledge of physics to be robust and functional. That
is, they should be able to recognize contexts that are appropriate for their particular physics
knowledge and use them in correct ways. This means that we need to help students not only
to obtain physics content knowledge, but to organize their knowledge.
These goals suggest that we should broaden our evaluation procedures. Traditionally we
only test the content and part of the student’s skill in doing physics, usually (at least at the
introductory level) in limited preset contexts. Sagredo once decided that an effective way to
get students to do their homework was to create exams made up of only previously assigned
homework problems. The students were quite happy with this arrangement. Unfortunately,
this sends the strong message that the only physics knowledge they need is restricted to a
small set of problems whose solutions can be memorized. Students trained in this fashion are
unlikely to be able to solve any other problems than the ones they have memorized. (I have
seen situations where reversing the figure from left to right made the problem impossible for
students who were quite comfortable solving the original problem.)
It is not sufficient for students to “know” the relevant correct statements of physics. They
also have to be able to gain access to them at the appropriate times; and they have to have
methods of cross-checking and evaluating to be certain that the result they have called up is
truly relevant. To do this, they need to build a coherent and consistent mental model of the
subject.
If we want to help our students build good models of the physics content, our cognitive
model of student learning provides some guidance. The experience that outstanding teachers
have reported is consistent with what we learn from researchers in cognitive and neuroscience:
activities teaching thinking and reasoning have to be repeated in different contexts over a pe-
riod of time. Arnold Arons observes:
It must be emphasized, however, that repetition is an absolute essential feature of [effective] in-
struction—repetition not with the same exercises or in the same context but in continually al-
tered and enriched context. . . . Experience . . . must be spread out over weeks and months and
must be returned to in new contexts after the germination made possible by elapsed time. Start-
ing with only a few students being successful, each repetition or re-cycling sees an additional per-
centage of the class achieving success, usually with a leveling of somewhat below 100% of the
total after approximately five cycles. [Arons 1983]
Our cognitive colleagues tell us that to move something from short-term to long-term
memory takes repetition and rehearsal. Our neuroscience colleagues are showing us that
learning involves real structural changes among neural connections. I summarize this with a
basic teaching and learning precept—really a corollary to the constructivism principle.
Corollary 1.1: Learning is a growth, not a transfer. It takes repetition, reflection, and in-
tegration to build robust, functional knowledge.
This corollary leads to a guideline for instruction. I refer to it (and to subsequent guide-
lines for instruction) as a “teaching commandment.” The full set (along with the cognitive
principles and goals listed in this chapter and elsewhere) are summarized in a file on the Re-
source CD.
Redish’s first teaching commandment: Building functional scientific mental models does
not occur spontaneously for most students. They have to carry out repeated and varied
activities that help build coherence.
CHAPTER 3
There’s More Than Content

to a Physics Course:
The Hidden Curriculum1
Education is what survives

when what has been learned
has been forgotten.
B. F. Skinner
(New Scientist, 21 May 1964)
In the last chapter, we discussed how our cognitive model of student thinking helps us
understand the importance of the ideas our students bring into the classroom. But cog-
nition is complex. Students do not only bring ideas about how the physical world works
into our classrooms. They also bring ideas about the nature of learning, the nature of
science, and what it is they think they are expected to do in our class. In addition, they
have their own motivations for success. We are often frustrated by the unspoken goal
of many of our students to be “efficient”—to achieve a satisfactory grade with the least
possible effort—often with a severe undetected penalty on how much they learn.
Most of my students expect that all they have to do to learn physics is read their text-
books and listen to lectures. Although some students who believe this don’t actually carry out
this minimal activity, even those who do often fail to make sense of physics in the way I want
them to. This leads me to believe that reading textbooks and listening to lectures is a poor
way of learning for most students. Sagredo objects, “This is clearly not universally true!” Re-
membering Principle 4, I concur. As physics teachers, most of us have had the experience of
having a few “good” students in our lectures—students for whom listening to a lecture is an
active process—a mental dialog between themselves and the teacher. Indeed, many of us have
1 This chapter is based in part on a paper by Redish, Saul, and Steinberg [Redish 1998].
51
52 • Chapter 3: There’s More Than Content to a Physics Course: The Hidden Curriculum
been that good student, and we remember lectures (at least some of them) as significant parts
of our learning experience.2
A similar statement can be made about texts. I remember with pleasure working through
some of my texts and lecture notes, reorganizing the material, filling in steps, and posing
questions for myself to answer. Yet few of my students seem to know how to do this or even
that this is what I expect them to do. This leads us to think about an additional observation.
Many of our students do not have appropriate mental models for what it means to learn physics.
This is a “meta” issue. People build schemas not only for content but also for how to
learn and what actions are appropriate under what circumstances. Most of our students don’t
know what you and I mean by “doing” science or what we expect them to do. Unfortunately,
the most common mental model for learning science in my classes seems to be:
• Write down every equation or law the teacher puts on the board that is also in the book.
• Memorize these, together with the list of formulas at the end of each chapter.
• Do enough homework and end-of-the-chapter problems to recognize which formula is
to be applied to which problem.
• Pass the exam by selecting the correct formulas for the problems on the exam.
• Erase all information from your brain after the exam to make room for the next set of
materials.
I call the bulleted list above “the dead leaves model.” It’s as if physics were a collection of
equations on fallen leaves. One might hold s 1⁄ 2gt 2, another F ma, and a third F kx.
Each of these equations is considered to have equivalent weight, importance, and structure.
The only thing one needs to do when solving a problem is to flip through one’s collection
of leaves until one finds the appropriate equation. I would much prefer to have my students
see physics as a living tree.
A SECOND COGNITIVE LEVEL

The issues discussed in the introduction to this chapter seem to be at a different level of cog-
nition than the more specific cognitive responses discussed in chapter 2. A number of
cognitive researchers have identified a second level of cognition that resides “above” and con-
trols the functioning of the level described in chapter 2 [Baddeley 1998] [Shallice 1988]
[Anderson 1999]. Many of them refer to this as executive function—thinking processes that
manage and control other thinking processes. In the context of instruction, three types of
cognitive controls are particularly important: expectations, metacognition, and affect.
Each student, based on his or her own experiences, brings to the physics class a set of
attitudes, beliefs, and assumptions about what sorts of things they will learn, what skills will
be required, what they will be expected to do, and what kind of arguments and reasoning
2 However, compare my discussion of one of my lecture experiences in the section of chapter 7 on “The Traditional
Lecture.”
Expectations: Controlling Cognition • 53
they are allowed to use in the various environments found in a physics class. In addition,
their view of the nature of scientific information affects how they interpret what they hear. I
use the phrase expectations to cover this rich set of understandings that are particular to a
given class. Metacognition refers to the self-referential part of cognition—thinking about think-
ing. Sometimes these responses are conscious (“Wait a minute. Those two statements can’t be
consistent.”), but the term is also used to refer to the unconscious sense of confidence about
thinking (“It just feels right.”). Under affect, I lump together a variety of emotional responses
including motivation, self-image, and emotion.
EXPECTATIONS: CONTROLLING COGNITION

Expectations affect what students listen to and what they ignore in the firehose of informa-
tion provided during a typical course by professor, teaching assistant, laboratory, and text.
They affect which activities students select in constructing their own knowledge base and in
building their own understanding of the course material. The impact can be particularly strong
when there is a large gap between what the students expect to do and what the instructor ex-
pects them to do.
Most physics instructors have expectation-related goals for their students, although we
don’t often articulate them. In our college and university physics courses for engineers, biol-
ogists, and other scientists, we try to get students to make connections, understand the lim-
itations and conditions on the applicability of equations, build their physical intuition, bring
their personal experience to bear on their problem solving, and see connections between class-
room physics and the real world. Above all, we expect students to be making sense of what
they are learning. I refer to learning goals like these—goals not listed in the course’s syllabus
or the textbook’s table of contents—as part of the course’s hidden curriculum.3
Expectations about learning

Students’ expectations about how and what they will learn in science classes have been stud-
ied all across the curriculum. For pre-college, many studies have demonstrated that students
often have misconceptions both about the nature of scientific knowledge and about what they
should be doing in a science class.4 Other studies indicate some of the critical items that make
up the relevant elements of a student’s system of expectations and beliefs. I focus here on
studies at the college and secondary levels.
Two important large-scale studies that concern the general cognitive expectations of adult
learners are those of Perry and Belenky et al. [Perry 1970] [Belenky 1986]. Perry tracked the
attitudes of Harvard and Radcliffe students throughout their college careers. Belenky et al.
tracked the views of women in a variety of social and economic circumstances. Both studies
found evolution in the expectations of their subjects, especially in their attitudes about knowl-
edge.5 Both studies found their young adult subjects frequently starting in a binary or received
3 The first use of this term that I know of is [Lin 1982].

4 See,for example, [Carey 1989], [Linn 1991], and [Songer 1991].
5 This brief summary is an oversimplification of a complex and sophisticated set of stages proposed in each study.
knowledge stage in which they expected everything to be true or false, good or evil, etc., and
in which they expected to learn “the truth” from authorities. Both studies observed their sub-
jects moving through a relativist or subjective stage (nothing is true or good, every view has
equal value) to a consciously constructivist stage. In this last, most sophisticated stage, the sub-
jects accepted that nothing can be perfectly known, and they accepted their own personal
role in deciding what views were most likely to be productive and useful for them.
Although these studies both focused on areas other than science,6 Sagredo and I both
recognize a binary stage, in which students just want to be told the “right” answers, and a
constructivist stage, in which students take charge of building their own understanding.7 Con-
sciously constructivist students carry out their own evaluation of an approach, equation, or
result, and understand both the conditions of validity and the relation to fundamental phys-
ical principles. Students who want to become creative scientists will have to move from the
binary to the constructivist stage at some point in their education.
An excellent introduction to the cognitive issues involved in student expectations is given
by Reif and Larkin, who compare the intellectual domains of spontaneous cognitive activi-
ties that occur naturally in everyday life with those required for learning science [Reif 1991].
They point out that an important component of executive function is “deciding when enough
is enough.” They note that knowledge in the everyday domain is very much about satisfic-
ing rather than optimizing.8 The kind of consistency, precision, and generality of principles
typical of scientific knowledge is neither necessary nor common in people’s everyday activi-
ties. Students often apply everyday-domain thinking when we want them to apply scientific-
domain thinking.
The structure of student expectations: The Hammer variables

In order to get a handle on the complex issues of executive control and expectations, we need
to begin defining specific characteristics so that we can talk about them and begin to think
about ways to further them with instruction. In a series of interesting papers, David Hammer
has begun this task [Hammer 1996a] [Hammer 1996b] [Hammer 1997]. In these papers, he
identifies a number of parameters that arise from the expectations that a student brings into
the physics class. Hammer’s three variables are listed in Table 3.1.
I refer to these attitudes as favorable or unfavorable, since to make reasonable progress
toward becoming a scientist or engineer, a student will find unfavorable attitudes limiting
and will have to make a transition to the attitudes listed in the favorable column.
Sagredo complains, “I certainly expect my students to have the attitudes that you call fa-
vorable when they enter my class. If they didn’t learn these attitudes in school, what can I do
about it?” One of the problems, Sagredo, is that we often actually encourage unfavorable at-
titudes without really being aware of it. While working on his dissertation, Hammer did a
case study with two students in algebra-based physics at Berkeley who were carefully matched
6 Perry specifically excludes science as “the place where they do have answers.”
7 Inmy experience true relativism is rare, but not unheard of, among physics students.
8 The term “satisfice” was introduced into economics and cognitive science by Herbert Simon, who won a Nobel
Prize for the work. The point is that in real-world situations people do not optimize. It takes too much effort. Rather,
they tend to seek an answer that is “good enough,” that is, one that is both “satisfactory and suffices.” This creates
implications for the variational principles that economists construct.
TABLE 3.1 The “Hammer Variables” Describing Student Expectations

[Hammer 1996a].
Favorable Unfavorable
Independence takes responsibility for takes what is given by authorities
constructing own (teacher, text) without
understanding evaluation
Coherence believes physics needs to be believes physics can be treated as
considered as a connected, unrelated facts or independent
consistent framework “pieces”
Concepts stresses understanding of focuses on memorizing and using
the underlying ideas and formulas without interpretation
concepts or “sense-making”
as to grade point average, SAT scores, etc., but who had decidedly different approaches to
learning physics [Hammer 1989]. The first student tried to make sense of the material and
integrate it with her intuitions. She didn’t like what she called “theory” by which she said,
It means formulas . . . let’s use this formula because it has the right variable, instead of saying,
OK, we want to know how fast the ball goes in this direction. . . . I’d rather know why for real.
The second student was not interested in making sense of what she was learning. For her,
the physics was just the set of formulas and facts based on the authority of the instructor and
text. Consistency or sense-making had little relevance.
I look at all those formulas, say I have velocity, time, and acceleration, and I need to find dis-
tance, so maybe I would use a formula that would have those four things.
Student A was able to make sense of the material for the first few weeks. Soon, however,
she became frustrated, finding it difficult to reconcile different parts of the formalism with
each other and with her intuition. Eventually, she compromised her standards in order to
succeed. Student B’s failure to seek consistency or understanding did not hurt her in the
course.
This small example indicates that we may inadvertently wind up encouraging students
to hold unfavorable attitudes. After learning about these issues, I tried to change the way I
taught in order to change this situation. How one might do this is discussed in chapter 4 on
homework and testing and in chapter 5 on surveys and assessing our instruction. I used the
Maryland Physics Expectations Survey (MPEX) we developed to test student expectations
(described in chapter 5 and given on the Resource CD). Although at first I didn’t get im-
provement, I learned that at least my grades were somewhat correlated with the results on
my survey, whereas those of my colleagues were not. This can be taken in two ways: Either
my survey is not measuring something we want students to learn, or our grades are not mea-
suring those behaviors we want to encourage.
As we begin to develop a more complex view of what is going on in a physics class, what
we want the students to get out of it, and what we value, we begin to realize that sometimes
“the right answer” is not the only thing we should be looking for. A dramatic demonstration
of student variability on attitudinal issues and how these issues play out in a classroom set-
ting is given by Hammer’s analysis of a discussion among a teacher and a group of high school
students trying to decide whether a ball rolling on a level plane would keep moving at a con-
stant speed [Hammer 1996a]. The students had been told Galileo’s arguments that under
ideal conditions it would do so.9 I’ve numbered the lines in the discussion so that we can re-
fer to them later.
1. Prior to this moment, the debate had mostly focused on the question of whether
it is friction, gravity, or both that causes the ball to slow down. The students also
debated whether it is appropriate to neglect friction or gravity, or both, and
whether it is possible to neglect one without neglecting the other.
2. About 20 minutes into the debate, Ning argued that Galileo’s ideal conditions
would mean no forces on the ball, including no friction and no gravity; and, she
claimed,“if you don’t put any force on it, it’s going to stay still or go at constant speed.”
Bruce elaborated on Ning’s statement, adding that there must be a force to make
the ball move:
3. Bruce: If there is no gravity and no friction, and there is a force that’s making it
move, it’s just going to go in a straight line at a constant speed. . . . What’s mak-
ing the ball move?
4. Amelia [over several other voices]: The forces behind it.
5. Susan: He [Galileo] said there was no force.
6. Bruce: If there’s no force pulling it down, and no force slowing it down, it would
just stay straight.
7. Harry: The ball wouldn’t move.
8. Jack: There’s no force that’s making it go.
9. Steve: The force that’s pushing it.
10. Bruce: The force that’s pushing it will make it go.
11. Jack: Where’d that force come from, because you don’t have any force.
12. Steve: No there is force, the force that’s pushing it, but no other force that’s slow-
ing it down.
13. Many voices at once, unintelligible. Sean says he has an example.
14. Teacher: Sean, go ahead with your example.
15. Sean: If you’re in outer space and you push something, it’s not going to stop un-
less you stop it.
16. Teacher: If you’re in outer space and you give something a push, so there’s a place
with no gravity—
17. Sean: No gravity, no friction.
18. Teacher: —it’s not going to stop until you stop it. So Penny what do you think
about that?
9 Student names are pseudonyms.

19. Penny: But we talked about the ball on [a surface], but when we talk about space,
it’s nothing like space. So I was just saying that gravity will make it stop.
20. Amelia objected to Sean’s example for another reason, saying that something
moving in space will still stop.
21. Amelia: No. Maybe there’s no gravity and no air there, but there are other kinds
of gases that will stop it.
22. Teacher: But those are other, those are outside things.
23. Amelia: The outside friction should stop it.
24. Bruce: That’s not, that makes it an un-ideal state.
25. Scott: Space is a vacuum. Like a vacuum there’s no—
26. Amelia: There are other kinds of gases.
27. [Several voices, unintelligible.]
28. Harry: We’re talking about ideal space. (students laugh)
29. I intervened at this point to steer the discussion away from the question of
whether there are gases in space and toward the question of whether there is
a “force that’s moving” the ball.
30. Teacher: . . . So how can one side say there are no forces on it, and the other
side say there is a force that’s moving it.
31. Bruce: Well there was an initial force.
32. Susan: There’s an initial force that makes it start, giving it the energy to move.
In analyzing this discussion, Hammer identifies half a dozen perspectives that could be
used to evaluate the students’ responses. I want to focus on four.
• Content answer: Does the student have the correct answer?

• Reasoning: Does the student display a common naïve conception? Is it related to a rea-
soning primitive?
• Coherence: Does the student understand that scientific laws are developed to unify a wide
variety of circumstances and that science should be consistent?
• Understanding idealizations: Can the student see the relevance of idealized or limiting
conditions?
In the dialog, Ning gave the correct answer (line 2) but did not participate in defend-
ing it. The discussion revealed that many of the students had the common naïve conception
represented by the facet “motion is caused by force” (lines 3, 8, 11, 12). Almost all of the
discussion was by claim and counterclaim without citing reasoning or evidence. The discus-
sion in lines 15–19 shows a distinction between Sean, who is trying to make a link between
two rather different physical situations, and Penny, who wants to keep them separate. This
can be interpreted as a difference in their understanding of the need for coherence in science.
Sean’s claim in line 15 tried to take the analysis to an idealized situation, without gravity or
friction. Amelia (lines 23 and 26) did not appear to be comfortable in thinking about the
simplified example.
In other examples cited by Hammer, students gave the correct answer to a problem, but
argued its validity by citing the text or teacher and being unwilling to think about the issue
for themselves.
These examples illustrate the complexity of our hidden curriculum and show how we
can begin to think both about what the student is bringing in to our classes and what the
student can gain from our classes in a more sophisticated way than just “are they giving the
right or wrong answers.”
Connecting to the real world

Although physicists believe they are learning about the real world when they study physics,
the context dependence of cognitive responses (see chapter 2) opens another possible gap be-
tween faculty and students. Most students seem to believe that physics is related to the real
world in principle, but a significant fraction also believe that what they are learning in a
physics class has little or no relevance to their personal experience. This can cause problems
that are both serious and surprising.
Even if our students develop strong concepts related to real-world meanings, the strong
context dependence of the cognitive response makes it particularly easy for students to re-
strict their learning in physics classes to the context of a physics class. This seems unnatural
to Sagredo. “Practically every problem I assign for homework or do on the board involves
some real-world physical context.” True, Sagredo. But that doesn’t mean that students will
easily or naturally make the connections that you do.
When an instructor produces a demonstration that has been “cleaned” of distracting el-
ements such as friction and air resistance, the instructor may see it as displaying a general
physical law that is present in the everyday world but that lies “hidden” beneath distracting
factors. The student, on the other hand, may believe that the complex apparatus is required
to produce the phenomenon and that it does not occur naturally in the everyday world, or
is irrelevant to it. A failure to make a link to experience can lead to problems not just be-
cause physics instructors want students to make strong connections between their real-life
experiences and what they learn in the classroom, but because learning tends to be more ef-
fective and robust when linked to real and personal experiences.
Even worse, students’ failure to connect their personal experience to what is happening
in their physics class can put up barriers to understanding that grow increasingly impenetra-
ble. As discussed in chapter 2, multiple representations are used in physics in order to code
knowledge in a variety of interlocking ways. A critical element in all of them is the map to
the physical system. An essential part of solving a problem is understanding what the real-
world version of the problem is, what’s important in that situation, and how it maps onto
physical principles and equations. If students don’t understand that part of the process, they
can have great difficulty in seeing the physics as a way to make sense of the physical world.10
10 The Physics Education Group at the University of Massachusetts—Amherst has done interesting research using
problem posing as a technique to help students develop these skills [Mestre 1991]. See also the variety of problems
discussed in chapter 4.
A shepherd has 125 sheep and 5 dogs. How old is the shepherd?
Figure 3.1 A word problem for middle school math students.
A classic word problem that illustrates this difficulty is shown in Figure 3.1. Although
this problem is patently absurd and cannot be answered, some middle school students will
struggle to find an answer (Expectation: “The teacher wouldn’t give me a problem that has
no solution.”) and will come up with an answer of 25. (“There are only two numbers to work
with: 5 and 125. Adding, multiplying, and subtracting them doesn’t give something that could
be an age. Only dividing gives a plausible number.”)
Another example comes from the mathematics exam given by the National Assessment
of Educational Progress (NAEP). A national sample of 45,000 13-year-olds was given the
problem shown in Figure 3.2 [Carpenter 1983]. Although 70% of the students who worked
the problem carried out the long division correctly, only 23% gave the correct answer—32.
The answer “31 remainder 12” was given by 29%, and the answer 31 was given by another
18% of those doing the problem. Thus, nearly half of the students who were able to carry
out the formal manipulations correctly failed to perform the last simple step required by the
problem: to think about what the answer meant in terms of a real-world situation. (Expec-
tation: “The mathematical manipulation is what’s important and what is being tested.”)
In these two examples, students are making somewhat different errors. In the shepherd
problem they are using some real-world information—what ages are plausible as answers; but
they are not asking how the numbers they are given could relate to the answer. They are not
making sense of the problem. In the soldiers and buses problem, students are not using their
real-world knowledge that you cannot rent a fraction of a bus. In both cases, students who
make these errors focus on the mathematical manipulations and fail to “make sense” of the
problem in real-world terms.
The same problems occur frequently in introductory physics. In my experience with in-
troductory college physics, more than half of the students do not spontaneously connect what
they learn in their physics class to their everyday experiences—either by bringing their every-
day experiences into their physics classes or by seeing the physics they are learning in the out-
side world. Two anecdotal examples show how this plays out in a college physics class.
A student in my algebra-based physics class missed a midsemester exam due to an ill-
ness, and I agreed to give her a makeup. One of the problems on the exam was the follow-
ing. “A high jumper jumps so his center of gravity rises 4 feet before he falls back to the
ground. With what speed did he leave the ground?” This is a typical projectile problem. My
An army bus holds 36 soldiers. If 1128 soldiers are being bused to their training
site, how many buses are needed?
Figure 3.2 A problem for the NAEP math exam for middle school students.
student knew the formula and punched the numbers into her calculator. When she handed
in her test and I looked over her answers, she had come up with the answer 7840 feet/
second. (Can you guess what she had done wrong on her calculator?) I asked her whether
her answer to that problem had bothered her. She shrugged and said, “That’s what the for-
mula gave me.” She saw absolutely no need to check her answer against her experience—and
incidentally, it had never entered her mind that she might have misremembered the formula,
incorrectly recalled the value of a parameter, or made an error in pressing the calculator keys.
This overconfidence in their memory and processing is a symptom I have seen in very many
students. They assume anything they remember must be correct.
A second example occurred in my engineering (calculus-based) physics class. For many
years now, I have been requiring estimation (Fermi-type) problems in my classes.11 Almost
every homework assignment has one, and every exam is guaranteed to have one. One of my
students came into my office hours and complained that this wasn’t fair. “I don’t know how
big these things are,” she scowled. “Well,” I said. “How about a foot? Do you know how big
a foot is?” “I have no idea,” she replied. Assuming that she was overstating her case to make
her point, I said, “How about making a guess? Show me how far up from the floor a foot
would be.” She placed her hand at about waist level. “And how tall are you?” I asked. She
thought for a second, said “Oh” and lowered her hand somewhat. She thought again and
lowered her hand again—to about the right height above the ground. She looked at her
hand—and at her foot a few inches away and remarked with great (and what appeared to be
genuine) surprise, “Oh! Does it have anything to do with a person’s foot?”
Since these real-world connections are critically important in developing an under-
standing of how physics helps us to make sense of our everyday experiences,12 I specify a
fourth learning goal.
Goal 4: Reality Link—Our students should connect the physics they are learning with
their experiences in the physical world.
To what extent does a traditional course help our students reach this goal? The simplest
way to find out is to ask them.13 In our study of student expectations in a calculus-based
physics class for engineers [Redish 1998], using the MPEX survey14 we found that student
expectations of the connection between physics and the real world typically tended to dete-
riorate as a result of the first semester of instruction.
The four items of the MPEX reality cluster are shown in Table 3.2. They ask whether
students expect to/have needed to15 make the link to their outside experiences for the class
and whether students expect to/have found that what they learn in physics can be seen in
11 For examples of these types of problems, see chapter 4 and the sample problems on the Resource CD.
12 This is especially true for our service students in engineering and biology.
13 This method is not very accurate since students often do not reflect and do not necessarily know how they think.
A better approach is to watch them solving problems alone or in a group using think-aloud protocols [Ericson 1993].
14 See chapter 5 for a detailed discussion of the MPEX. The full survey and instructions on its use are contained in
the Action Research Kit on the Resource CD.

15 The alternate forms are for the pre- and post-class surveys.
TABLE 3.2 Results on the MPEX Reality Link Cluster Items

Favorable Unfavorable Favorable Unfavorable
MPEX Item Pre Pre Post Post
Physical laws have little relation 84% 5% 87% 2%
to what I experience in the
real world. ()
To understand physics, I 59% 11% 54% 22%
sometimes think about my
personal experiences and
relate them to the topic
being analyzed. ()
Physics is related to the real 73% 9% 61% 19%
world, and it sometimes
helps to think about the
connection, but it is rarely
essential for what I have to
do in this course. ()
Learning physics helps me 72% 10% 51% 18%
understand situations in my
everyday life. ()
their real-world experiences. Both issues are addressed in two statements, one positive and
one negative. The student’s response is considered to be favorable if she sees the need for a
connection and unfavorable if she does not. The polarity of the favorable result is indicated
after the item by a () when the favorable result is agree and by a () when the favorable
result is disagree. The students are asked to report on a five-point scale (strongly agree, agree,
neutral, disagree, strongly disagree), but for a favorable/unfavorable analysis, we ignore whether
or not there is a “strongly.” The responses come from pre- and post-surveys given in the first
semester of an engineering physics class. The class was calculus-based and covered mostly
Newtonian mechanics. The results are shown for N 111 students (matched, i.e., who com-
pleted both pre- and post-surveys).16
The results are discouraging, especially on the last two items. I tried to help my students
make the connection by giving some estimation problems, but that was clearly insufficient.
Similar results have been found with other faculty teaching this class at Maryland and at many
other colleges and universities [Redish 1998].
There has been little published work on how to help students develop a strong reality
link. In my experience, regular essay questions asking the students to relate the physics to
their experience and regular estimation questions (being sure to include both on every exam
so that students take them seriously) only help a little bit. Even in lessons where physicists
see real-world implications immediately, students rarely make the connections spontaneously
16 A total of 158 students completed the class.

if not led to them. I expect this goal will only be achieved by a thorough interweaving of the
physics with explicit connections to the students’ experience.17 Further research and devel-
opment on this issue would be most welcome.
METACOGNITION: THINKING ABOUT THINKING

The transcript from David Hammer’s high school class in our earlier discussion shows that
different students access different kinds of reasoning in their discussion of a physics problem.
This variety arises from students having different expectations about the nature of science
and what it means to learn science. Unfortunately, many of these expectations are inappro-
priate for learning science. They may be learned in school, from movies and TV, or from
reading science fiction books.18 When students have the wrong expectations about what they
are supposed to do in a class, those expectations can serve as a filter, causing them to ignore
even explicit instructions given by the instructor.
In part, the approaches to learning physics that students bring into our classes arise from
a misunderstanding of the nature of scientific knowledge and how one has to learn it. As
pointed out so clearly by diSessa and discussed in chapter 2, for most ordinary people (even
for some of our best students19) knowledge of the world comes in “pieces” about how par-
ticular situations work [diSessa 1993] [diSessa 1988]. As pointed out by Reif and Larkin [Reif
1991], building a consistent and economical set of principles—at the cost of requiring long
and indirect explanations of many phenomena—is not the way most people create their mod-
els of the physical world in their everyday lives. It seems that people tend to look for quick
and direct explanations. The complex consistent and parsimonious net of links built by sci-
ence is not a natural type of mental construction for most people. It has to be learned.
The key element in the mental model I want my students to use in learning physics ap-
pears to me to be reflection—thinking about their own thinking. This includes a variety of
activities, including evaluating their ideas, checking them against experience, thinking about
consistency, deciding what’s fundamental that they need to keep and what is peripheral and
easily reconstructed, considering what other ideas might be possible, and so on. My experi-
ence with students in introductory classes—even advanced students20—is that they rarely ex-
pect to think about their knowledge in these ways. Students often come to my office hours
for help with problems. I always ask them to show me what they have tried so far and pro-
ceed to offer help via questions. They frequently have an error close to the start of their
analysis—in a principle or equation that they bring up from their memory. As I lead them
to implausible and unlikely results through my questioning they become troubled, but they
are much more likely to try to justify a ridiculous result by difficult and inconvenient con-
torted reasoning than by asking if one of their assumptions might be wrong.
17 Preliminary results with a more synergistic approach appear quite favorable [Redish 2001].
18 Some science fiction books, especially those written by scientists (such as David Brin, Gregory Benford, or John
Kramer), have excellent descriptions of the way science develops its knowledge.
19 Recall that in [diSessa 1993] the subjects studied were MIT freshmen.
20 Many of the students in my algebra-based physics classes are upper division students who have previously taken
many science classes in chemistry and biology.

Metacognition: Thinking About Thinking • 63
From our cognitive model we understand that to create new, coherent, and well-
structured mental models, students need to go through a number of well-designed activities
addressing the issue to be learned, to repeat them, and to reflect on them. Similar principles
hold for metacognition—thinking that reflects on the thinking process itself. I add another
learning goal to the list developed in chapter 2.
Goal 5: Metalearning—Our students should develop a good understanding of what it

means to learn science and what they need to do to learn it. In particular, they need to
learn to evaluate and structure their knowledge.
This is not a trivial goal and it does not happen automatically for most students as they work
to learn physics content.
Redish’s second teaching commandment: In order for most students to learn how to learn
and think about physics, they have to be provided with explicit instruction that allows
them to explore and develop more sophisticated schemas for learning.
“Hold on!” Sagredo complains. “I never have time enough to teach all the content I’m
supposed to teach. How can I find time to give them lessons in how to learn?” I sympathize,
Sagredo. But in fact, the problem is not as bad as it looks. If we are teaching them to learn,
we have to be teaching them to learn something. That something can easily be the appropri-
ate physics content. Some introductory discussion, lessons designed to encourage particular
activities, and reflections analyzing what they’ve done should help substantially. One of the
few well-documented approaches to explicitly teaching and improving students’ metacogni-
tion is the work of Alan Schoenfeld.
Instructional techniques for improving metacognition

Alan Schoenfeld, in a problem-solving college math class, developed a group-problem-
solving method that focused on helping students strengthen their judgment and control of
their own thinking. The class was small enough (on the order or fewer than 25 students) that
he could use a guided cooperative group-problem-solving approach.21
In his observations of the class’s behavior, Schoenfeld found that his students often wasted
a lot of time in following unproductive approaches through a lack of metacognitive activity.
The students quickly jumped on the first idea that came to their minds and then proceeded
to “churn” through extensive manipulations, frequently losing track of what they were doing
and rarely evaluating whether their approach was productive.
Schoenfeld developed an instructional method to help students become more metacog-
nitively aware. The key was the mantra of metacognitive questions posted on the wall shown
in Figure 3.3. His comments on how this worked are worth repeating.
Students’ decision-making processes are usually covert and receive little attention. When students
fail to solve a problem, it may be hard to convince them that their failures may be due to bad
21 Seechapter 8 for a discussion of a method of this type employed in physics to help develop students’ conceptual
development and problem-solving skills.
What (exactly) are you doing?

(Can you describe it precisely?)
Why are you doing it?
(How does it fit into the solution?)
How does it help you?
(What will you do with the outcome when you get it?)
Figure 3.3 Schoenfeld’s questions for helping students learn to focus on metacognitive issues.
decision-making rather than to a lack of knowledge. The instructor had the right to stop stu-
dents at any time while they were working on the problems and to ask them to answer the three
questions on [Figure 3.4]. At the beginning of the course the students were unable to answer the
questions, and they were embarrassed by that fact. They began to discuss the questions in order
to protect themselves against further embarrassment. By the middle of the term, asking the ques-
tions of themselves (not formally, of course) had become habitual behavior for some of the stu-
dents. [Schoenfeld 1985]
Schoenfeld not only implemented a focus on metacognition and control in the group
activity, but he modeled it in his approach to modeling solutions for the class as a whole. His
description outlines the process in detail.
When the class convened as a whole to work problems (40–50% of class time), I served as or-
chestrator of the students’ suggestions. My role was not to lead the students to a predetermined
solution, . . . my task was to role model competent control behavior—to raise the questions and
model the decision-making processes that would help them to make the most of what they know.
Discussions started with “What do you think we should do?” to which some student usually
Stages, time spent on each stage, sequencing and manage- Stages, time spent on each stage, sequencing and manage-
Episode or ment activity by Novices KW and AM in Protocol 9.2. Episode or ment activity by Semi-Novices KW and DR in Protocol 9.4.
stage stage
Read Read
Analyze Analyze
Explore Explore Transition
Plan Plan
Implement Implement
Verify Verify
10 20 10 20
Elapsed time (minutes) Elapsed time (minutes)
Figure 3.4 Sample plots of student activities in solving math problems in Alan Schoenfeld’s metacognitive math class.
Small triangles mark metacognitive statements [Schoenfeld 1985].
Affect: Motivation, Self-Image, and Emotion • 65
suggested “Let’s do X.” Often the suggestion came too rapidly, indicating that the student had
not adequately thought through what the problem called for or how the suggestion might be use-
ful. The class was then asked, “Are you all sure you understand the problem, before we proceed
with X?” A negative response from some students would result in our taking a closer look at the
problem. After doing so, we returned to X as a possible solution approach. Did X still seem rea-
sonable? Not infrequently the answer was “no.” When it was, this provided the opportunity to re-
mind students about the importance of making sure that one has understood a problem before
jumping into its solution. . . . After a few minutes of working on the problem—whether or not
we were on a track that would lead to a solution—the process would be halted for an assessment
of how things were going. The class was asked “We’ve been doing this for 5 minutes. Is it useful,
or should we switch to something else? (and why?)” Depending on the evaluation, we might or
might not decide to continue in that direction: we might decide to give it a few more minutes
before trying something else. Once we had arrived at a solution, I did a postmortem on the so-
lution. The purpose of that discussion was to summarize what the class had done and to point
out where it could have done something more efficiently, or perhaps to show how an idea that
the class had given up on could have been exploited to solve the problems. . . . The same prob-
lem was often solved three or four different ways before we were done with it. [Schoenfeld 1985]
By the end of the class, Schoenfeld found that the students were spending a much larger
fraction of their time in planning and evaluation and that their “metacognitive events” (state-
ments like: “I don’t understand this” or “That doesn’t seem right”) more often led to their
jumping into planning or checking mode than it did at the beginning of the class. This is il-
lustrated in Figure 3.4.
AFFECT: MOTIVATION, SELF-IMAGE, AND EMOTION

It is patently clear to most university physics instructors that motivation, how students feel
about the class, and how the students feel about themselves, play a critical role in how stu-
dents respond to instruction and how well they learn. The issues of feeling, emotion, and
mood are summarized by the term affect or affection in psychology. These issues have been
discussed extensively in the educational literature, [Graham 1996] [Stipek 1996], but I do
not attempt to review this literature here as the interaction between affect and cognition is
extremely complex and it is difficult to provide precise guidance. This is not to say these is-
sues are not of great importance. I therefore make a few comments, but refer the reader to
the literature cited above for details.
Motivation
Motivation can be a major factor in distinguishing students who will make the effort to learn
and those who will not. We encounter a variety of motivations.
• Internally motivated—Some students who come to our classes are self-motivated by an

interest in physics and a desire for learning.
• Externally motivated—Some students have no internal interest in physics but are strongly
motivated to get a good grade because our class is a hoop that must be jumped through
for them to get into a program for which they are motivated.
• Weakly motivated—These students are taking physics because it is a requirement, but

they are concerned only about passing, not getting a good grade.
• Negatively motivated—Some students are motivated to fail—for example, in order to
demonstrate to a controlling parent or mentor that they are not suited to be an engineer
or a doctor.
Those in the first group are a physics instructor’s delight. Whatever you give them they
make the most of. We can work with those in the second group by controlling the learning
environments we set up and making clear what will be evaluated on exams. (See examples in
chapter 4.) I can rarely do anything with the last group. Their goals in the class are distinctly
different from mine.
Finding ways to motivate your students to want to learn physics can be an extremely ef-
fective lever to improve the success of your teaching. Unfortunately, this is easier said than
done and is where much of the “art” in teaching comes in. It is easy to mistake student hap-
piness for student motivation. Making your lecture “entertaining” does not necessarily in-
crease students’ motivation for learning. Indeed, it can set up the expectation in their minds
that associates your lecture with a TV program where they don’t have to think.
Providing connections to their chosen career sometimes helps. I evolve my estimation
problems into design problems in my engineering physics class and create problems with a
medical and biological context for my algebra-based students. I hope this helps them see the
relevance of physics to a profession toward which they should, in principle, be motivated.
(Interviews with a small number of volunteers—usually the better students—suggest that at
least this group is making the connection [Lippmann 2001].)
Motivation is perhaps the primary place where the teacher in fact makes a significant
difference. A teacher with the empathy and charisma to motivate the students can create sub-
stantially more intellectual engagement than one who reads from the book and does not take
the time to interact with the students. Perhaps the most critical element in creating motiva-
tion is showing your students that you are interested in them, you want them to succeed,
and you believe that they can do it.
Self-image
Sagredo is a bit skeptical about the issue of students’ self-image. He feels that the education
community pushes “helping students feel good about themselves,” sometimes to the detri-
ment of serious critical self-analysis and learning, at least if the letters to the editor published
in newspapers are to be believed. In my experience with university-level physics students, this
issue cuts two ways. Some students are supremely overconfident, while others think that they
cannot possibly understand physics. Both groups are difficult to deal with.
In our small-group-sessions, we often use the Tutorial materials developed at the Uni-
versity of Washington. These lessons are research-based group-learning worksheets (see chap-
ter 8 for a detailed description) and use a cognitive conflict model. As a result, students who
are used to being right often feel the Tutorials are trivial and therefore useless—even when
they are consistently getting the wrong answers. When I am facilitating in one of these ses-
sions, I see this as a terrific learning opportunity. I circulate through the class, asking what
they got on the tricky questions. When I find a group that has been overwhelmed by an over-
confident student with a wrong answer, I say, “Now remember: Physics is not a democracy
and physics is not determined by charisma. You can’t tell who’s right by who says it the most
Affect: Motivation, Self-Image, and Emotion • 67
30 Women
Men
25
20
15
10
0
Gender threat No threat
Figure 3.5 Scores of college sophomore males and females on a math test when a comment is made
that the test “tends to separate genders” and when no such comment is made [Steele 1997].
forcefully or by what most people think. It has to make sense and it has to be consistent. Per-
haps you want to go back and think that question out again.” The result is almost always
that someone else in the group who had previously been intimidated into silence can bring
everyone to the correct result. This sends a really useful message—both for the overconfident
student and for the other members of the group.
On the other side, I have had experience with students who were absolutely convinced
that they were incapable of learning physics. In one case, I had a student in algebra-based
physics who was convinced “she couldn’t do this stuff ” and told me so repeatedly. However,
I often watched her vigorously argue difficult issues in Tutorials with another student who
was supremely confident of her ability and answers. My underconfident student was almost
always right, and my overconfident student almost always wrong.
Despite her success in Tutorials, this student did not change her overall self-evaluation
of her ability and she did poorly on exams. In other cases, I was able to help students who
were good in other classes but who, perhaps because of bad experiences in high school, were
convinced that they “couldn’t do physics.” All these cases are best treated carefully and indi-
vidually, using all the empathy and understanding you can bring to bear. Unfortunately, in
many college and university situations, the pressure of time and numbers makes it difficult,
if not impossible, to allow you to offer the individualized responses needed.
There has been some research on the topic of math anxiety or “math phobia.” (See, for
example, [Tobias 1995].) I do not know of comparable work on “science phobia.” There has
also been some extremely important work on the implications of social stereotypes on self-
image and performance. Stanford sociologist Claude Steele explored the implications of rais-
ing the link in a student’s mind to gender or race in conjunction with a mathematics test
[Steele 1997]. College sophomores who had committed themselves to a math major or mi-
nor were given a test somewhat above their level. One group was told that the test was “just
a trial” and that the researchers “wanted to see how they would do.” A second group was told
up front that the test “showed gender differences.” (The sign of the difference was not spec-
ified.) The results, shown in Figure 3.5, were dramatic. In the group given the test without
any comment about gender, males and females scored approximately the same. In the group
with the comment about gender, referred to as a gender threat by Steele,22 females scored sig-
nificantly worse (by more than a factor of 3!), and males scored somewhat better (about 50%).
The implication appears to be that stereotypes (males are better in math) pervade our
culture in a profound way, with implications that we tend to be unaware of and are insensi-
tive to. This certainly suggests that we should be extremely cautious about making any com-
ments at all about gender or race to our classes. For researchers, it suggests that in doing in-
terviews or surveys, questions about the respondent’s gender, race, or other social factors should
be given separately after the testing is complete.
Emotion
“I’m a physicist, not a song-and-dance-man!” Sagredo complains, echoing Star Trek’s Dr.
McCoy. Perhaps, Sagredo, but making your students feel good about your class can have an
influence on their learning. For one, if they hate your lectures and don’t come to class, they
won’t be able to learn anything from them.23 On the other hand, if you fill your lecture with
jokes, films, and cartoons, they are unlikely to take them seriously.
The best thing you can do to make students “feel good” about your class is to make it
worthwhile, at an appropriate level, and fair. Students like to feel that they are learning some-
thing valuable and that they can get a “good” grade (this may have different meanings for
different students) without having to work so hard that their other classes (and their social
lives) suffer. Getting students to learn a lot from our classes is a process of negotiation. From
my point of view as a teacher, I want them to work hard, but from their point of view as a
student, they don’t want to work hard without a clear payoff. In physics, learning can be frus-
trating and nonlinear. Often you have to work for a long time without feeling that you are
making much progress. Then, suddenly, everything falls into place and it all makes sense. But
until the “click,” you can’t be sure how much time you need to “get it” and it’s difficult to
plan. Students first have to learn what understanding the physics feels like and be slowly
drawn into working hard enough to learn harder and harder topics.
But entertainment and “song-and-dance” don’t have to be shunned, Sagredo. In our con-
text, it can mean little physics jokes, personalized stories, and dramatic demonstrations. (But
see the discussion of demonstrations in chapter 7.) All of these can be effective—or not. Jokes
should be relevant, not off-color, and not derogatory to groups or individuals. Personalized
stories should be relevant to the physics involved and have some point that will make sense
to a novice. They shouldn’t occupy so much of the time that students begin to feel you’re not
offering them enough physics. Demonstrations can be the best but are also dangerous. As ex-
plained in chapter 7, demonstrations can be entertaining but misleading. Students often don’t
see what you think they are seeing. A careful and involving class discussion, both before and
after the demonstration, is usually needed.
22 Note that the “threat” is implicit. There was no statement as to which group was expected to do better, and there
were no consequences for the students no matter how they scored.
23 Students tend to learn little even from lectures they attend unless special tools are used. See chapter 7.
CHAPTER 4
Extending Our Assessments:

Homework and Testing
If I had to reduce all of educational psychology

to just one principle, I would say this:
The most important single factor influencing learning
is what the student already knows.
Ascertain this and teach him accordingly.
D. P. Ausubel [Ausubel 1978]
The two-level cognitive model we develop in chapters 2 and 3 has implications for the
way we evaluate our students and our instruction. First, our understanding of the con-
text dependence of the cognitive response leads us to appreciate the importance of stu-
dents developing functionality and coherence in their knowledge, not just learning iso-
lated facts or memorizing a small set of problem solutions. Second, our understanding
of the associative character of the cognitive response leads us to appreciate the impor-
tance of building a good conceptual base and helping students link their conceptual un-
derstanding to their problem-solving and analysis skills. Third, our understanding of stu-
dent expectations and the stability of well-established mental models leads us to
appreciate that changing what our students value and what they choose to do to learn
is going to require a negotiation between instructor and students.
This negotiation has two important components: (1) We need to offer our students ac-
tivities that promote the kind of thinking we want to encourage (seeking coherence, sense-
making, effective analysis, creativity and so on), and (2) we need to set up mechanisms that
permit two-way feedback.
What students actually “do” in our classes usually consists of a variety of activities in-
cluding both in-class activities, such as listening to a lecture or doing a lab, and out-of-class
activities, such as reading the text or doing homework.1
1 Inthe education literature, activities that are done by all students together in a class are referred to as synchronous.
Activities that are done by students out of class independently at their own time and pace are called asynchronous.
69
70 • Chapter 4: Extending Our Assessments: Homework and Testing
In a negotiation, there must be feedback in both directions. We, the instructors, need to
know what our students are doing and how much they have learned. Students in our classes
need to know when they are meeting the goals we have for them and how to modify their
activity when things aren’t working.
In this chapter, I discuss the role of homework and testing in our classes. (Other com-
ponents of the class are discussed in chapters 7–9.) These activities play critical roles in help-
ing our students understand what it means to learn and do physics. I begin with a general
discussion of the goals of assessment and evaluation. Next I specifically discuss homework
and testing and present some of the methods I have found useful. I then review some of the
different kinds of questions we can use both in homework and on exams to help our stu-
dents understand what it is we want them to get out of our course and to help us under-
stand the extent to which they have done it.
ASSESSMENT AND EVALUATION

If we are interested in probing how well our instruction is working, we can think about the
answer along two axes, depending on our answer to the questions “when?” and “what?”
• When? Are we asking while the class is still in session or at the end?
• What? Are we asking about a student or about the class as a whole?
Probes of the success or failure of our instruction that occur during the class can help
both the student and the instructor correct things that are going wrong and identify prob-
lems that still have a chance to be fixed. Such probes are called formative. Probes that occur
at the end of a class and that serve to provide a final certification of student success are called
summative. There is also a difference in whether we are probing individual students or our
instruction as a whole. I refer to probes of an individual student’s learning as assessment and
to probes of our instruction as a whole as evaluation.
In this chapter, I focus on looking at assessment through the lens of the question, “What
has a particular student learned?” In the next chapter, I consider evaluation and the question,
“How effective was my instruction for the group of students in my class?” This separation is
somewhat artificial, since when we assess our individual students’ learning it tells us some-
thing about how successful our instruction has been overall. When we try to evaluate our in-
struction overall, the only way we can do it is by probing individual students. We will see,
however, that some tools that are useful for evaluation of an overall result may not do an ad-
equate job of assessing an individual’s learning.
To get feedback on how our class is doing and to measure the success of our interven-
tions, we must first decide what we mean by “success.” The discussions in chapters 2 and 3
show that this can be a complex issue. Our choice of goals plays an important role in deter-
mining our approach to assessment. What we mean by success is, in turn, determined by our
model of student understanding and learning.
To evaluate a student’s learning, we have to take account of a number of factors, in-
cluding conceptual learning, problem-solving skills, coherence and appropriateness of their
patterns of association/mental models, and so forth. We must recall that a student may “have”
an item of knowledge, that is, be able to recall it in response to a narrow range of cues, but
Homework • 71
be unable to recall and apply it in a wide range of appropriate circumstances. As a conse-

quence, our probes need to be varied and diverse. Furthermore, we need to watch out for
“wishful thinking” such as “filling in the blanks.” Often, a student will offer a statement or
an equation, and we will assume that they have come to that result in the same way that we
would have. Unfortunately, that is often not the case.
GIVING FEEDBACK TO YOUR STUDENTS

As we learned in chapters 2 and 3, students can have ways of looking at and analyzing physics
that are quite different from those expected by the teacher. Significant conceptual difficulties
may be buried in a student’s “wrong” answer to a homework or exam problem, and the stu-
dent may not have the knowledge or self-evaluation skills to disentangle them. Real feedback
on their thinking is immensely valuable to them. Unfortunately, this is one of the first things
to be cut when financial or personnel constraints get tight and classes get large. Homework
grading disappears in favor of only assigning problems for which the answer is given in the
book. Detailed solutions can help somewhat, but since there are often many ways to approach
a particular problem, they do not always help students debug their own thinking.
Redish’s third teaching commandment: One of the most useful aids you can give your stu-
dents is detailed feedback on their thinking—in an environment in which they will take
note of and make use of it.
HOMEWORK
Sagredo says that he always assigns homework in his physics classes, since it’s in doing the
problems that the students “really learn the physics.” Here’s a place where he and I strongly
agree. In my experience, far too many students have the expectation that physics is a set of
simple “facts” and a list of equations, which, if memorized, should permit them to get an A
in my class. Far too few are able to solve the complex kinds of problems that require them
to think deeply about the physics and to build a strong conceptual understanding and an ap-
propriate and well-organized knowledge structure. Homework can be of immense value in
helping students learn physics, but the wrong kind of homework can also send the wrong
message, confirming their view that “facts and equations are enough.”
Unfortunately, sometimes despite our best efforts, the wrong message still gets through.
End-of-chapter problems given in most introductory texts come in a variety of formats—
questions (often conceptual), exercises (straightforward plug-and-chug), and problems (more
complex reasoning required). Sagredo and I have both assigned a mix of problems for many
years. But recently, I have come to suspect that for many of our students this is still doing
them a disservice. A significant fraction of my students appear to think (or hope) that what
my course is really about is straightforward plug-and-chug. This is a pretty standard wishful
thinking, since these are the problems they can comfortably do. If they write something ar-
bitrary on the questions, get all the exercises right, and write down enough to get partial
credit on the problems, they are satisfied that they have gotten all that is necessary out of the
assignment. In fact, these students largely miss the main point. The questions are supposed
to get them to think deeply about the concepts, and the problems are supposed to help them
build a good understanding of how to work with physics. Given the “escape hatch” of the
exercises, some students make little effort to get any farther.
As a somewhat draconian attempt to close this loophole, I have stopped giving students
exercises as part of their homework. Instead, I reduce the total number of problems and give
a smaller number of harder problems, including essay questions and context-rich problems.
(See the explanation of these problems below.) One of my better students complained about
this. He said he needed the exercises to gain familiarity with the material. I responded that
he was perfectly welcome to do them and that many had answers in the back of the book so
he could check his work—but that success in this class means much more than plug-and-
chug problems and that he would be evaluated on what I wanted him to learn in the end.
Even when we give complex problems, we tend to “degrade” them in response to stu-
dent pressure. As described in chapter 2 in the section on multiple representations, a major
difficulty in solving a realistic physics problem is the extraction of a solvable physics prob-
lem from a complex real-world situation. When we assign variable names to the quantities
in the problem situation, when we only mention quantities that are relevant and give no ex-
traneous information, when we set up the problem so that all choices of what’s important
and what’s not are explicitly indicated, we steal from the student the opportunity to learn a
whole range of essential problem-solving skills.
When students find some tasks difficult, we have a tendency to give in to their pres-
sure and modify the problems so that there is no need for them to develop those skills. A
crucial example is the ability to manipulate and interpret equations containing many sym-
bols. Sagredo is aware that students have this difficulty and is disturbed by it. “Why should
it matter? A symbol just stands for a number anyway. They should be able to use equations
either way.” I agree, Sagredo, but recall that experience and familiarity with representations
and manipulations make a big difference in the ability to recognize appropriate actions.2 I
once looked carefully at a couple of standard introductory calculus texts. To my surprise I
found that very few equations anywhere in the text contained more than one symbol. Those
that did followed a strict convention: x, y, z, and t were variables, while a, b, and c were
constants. If we want students to learn the skill of working with multisymbol equations, we
need to provide examples that will make them use it. Problems in which numbers are given
right away allow the students to avoid having to learn the skill of interpreting equations
with multiple symbols.
I find that weekly homework using a variety of challenging problems (especially if some
are marked “from an hour exam” or “from a final exam”) can be effective in helping to re-
orient students’ ideas of what they need to do to learn physics.
In addition to providing a rich venue for student activity, homework also plays a criti-
cal role in providing students with formative feedback. In many large classes, we cannot give
quizzes and examinations frequently enough to provide students enough feedback to modify
a learning approach they may have developed in high school or in other science classes. As I
discussed in chapter 3, student expectations play a major role in what they do. Even if you
give them good feedback, there is no guarantee that they will use it. If homework problems
are graded with comments but handed back weeks after they were done, the students might
2 Recall, for example, the Wason card (K2A7) example discussed in chapter 2.
Getting Feedback From Your Students • 73
not be able to reconstruct their state of mind when they did the problems. Any difficulty
they have interpreting written feedback is likely to result in their ignoring it.
GETTING FEEDBACK FROM YOUR STUDENTS

The Ausubel quote in the epigraph, our cognitive model, and my experience all agree: Feed-
back needs to work both ways. You can improve your teaching substantially by getting good
and regular feedback about where your students are, what they are thinking, and how they
are interpreting the information provided in your course.
Redish’s fourth teaching commandment: Find out as much as you can about what your
students are thinking.
Four plausible and frequently used approaches to getting feedback from your students
and evaluating their learning are:
1. Observe student behavior in class and in office hours.

2. Measure student satisfaction with a survey or questionnaire.
3. Measure student learning using a closed-ended question (multiple-choice or short-
answer), designed using the results of physics education research on commonly found
errors to specify attractive distractors.
4. Measure student learning using open-ended (long-answer or essay) exam questions—
problems or open-expression questions in which students explain and discuss their
answers.
The first approach is an essential part of understanding our students and seeing how they
are responding to our instruction. Here, we have to be particularly careful not to go into
wishful-thinking mode and assume that a question that a student asks in office hours simply
needs to be answered in a direct and straightforward manner. This may sometimes be the
case, but more often, I have found that asking a few well-placed questions such as “Well,
what do you think is going on?” or “Why do you ask?” or even “Could you explain why
you’re stuck?” often produces a lot of information—and indicates that my “gut-response” an-
swer would have been useless.
Redish’s fifth teaching commandment: When students ask you a question or for help, don’t
answer right away. Ask them questions first, in order to see whether your assumptions
about their question are correct.
The second approach, an attitude survey or questionnaire, is the simplest and most com-
monly used, but although student satisfaction is important in motivating student work, and
presumably therefore in producing student success, the link between satisfaction and learn-
ing is highly indirect. Indeed, students whose primary goal is a good grade may find higher
satisfaction in a course that produces a good grade without improved learning, since improved
learning often requires time and painful effort.
The third method, multiple-choice questions with research-based distractors or short-an-

swer questions with research-motivated contexts, is easy to deliver, but, as explained in chap-
ter 5, requires a substantial effort to develop effectively. I discuss four kinds of such questions
below: multiple-choice with research-based distractors, multiple-choice multiple-response,
representation translation, and ranking tasks. Other formats for short-answer questions have
been developed and can be quite effective [Peterson 1989].
The fourth approach, long-answer or open-ended questions, is easy to deliver, but the
grading and analysis of long-answer questions and problems can be time consuming. Student
answers must be read in detail and evaluated by the understanding displayed. Grading here
can be quite subtle. There is a tension between grading that is too draconian and only gives
credit for detailed and careful reasoning, and grading that is too casual and gives points for
remembered equations or principles that the student has no idea how to use. The former
tends to produce grades that are too low and the latter to send the message that understanding
is not important. Later in this chapter, I discuss four kinds of long-answer questions: open-
ended or context-rich reasoning problems, estimation questions, qualitative questions, and
essay questions.
TESTING
Often the standard testing we carry out to assess our students is limited, especially in large
classes. In order to facilitate grading, we might choose to resort to machine-graded closed-
answer questions—multiple-choice end-of-chapter problems that rely primarily on recogniz-
ing the correct equation and performing some straightforward algebraic manipulations to get
a number. Decisions to test in this way have a powerful impact on our instruction. Not only
do we severely limit the kind of feedback we receive about how our students are doing, but
we send them a powerful message about what we believe is important for them to learn.
College students are as busy as their instructors, having many courses to consider, not
just physics. In addition, many of them are young adults seeking partners and engaging in
extramural sports or clubs, and some are working long hours to pay for their tuition. Some
of these extramural activities will have more impact on their lives and careers than their physics
class, hard as this might be to believe! As a result, even if I provide my students with good
advice—such as that working something out in multiple ways is important for their learn-
ing, or that once they have finished a problem they should think about it to see that it makes
sense—they are likely to ignore it if they can’t see how doing it will pay off directly in points
and grades. Most students regularly practice “time triage,” doing only those activities that
seem absolutely necessary for getting through the class successfully.3
In order to get students to engage in the kind of intellectual activity we want them to
engage in, we have to let them know by testing them for it directly. In particular, in choos-
ing exam questions, we have to be aware of the mixed messages we may be sending. This can
be quite difficult, especially in these days where a class’s success is measured by “student hap-
piness” as evaluated with end-of-class student anonymous comments, as opposed to being
measured by some assessment of student learning. Many students are satisfied with getting
3 And of course “successfully” means different things to different students. For a few it means showing their parents
that they were not cut out to be doctors or engineers or whatever goals their parents may have imposed on them.
Testing • 75
through a class with a decent grade without having to do too much work—even if they learn
little or nothing—and on end-of-semester surveys they may reward teachers who help them
achieve this goal and punish those who make them work too hard.4
Designing exams
The kinds of questions we choose for an exam can make a big impact on what students choose
to do in our classes. If, for example, we construct our exams from multiple-choice questions
that can be answered by recognition (that is, there are no “tricky” or tempting alternatives),
our students are likely to be satisfied with reading their text and lecture notes “lightly” a num-
ber of times. If we construct our exams from homework problems they have been assigned,
our students are likely to seek out correct solutions from friends or solution books and to
memorize them without making the effort to understand or make sense of them. If all our
exam problems can be solved by coming up with the “right equation” and turning the crank,
students will memorize lists of equations and pattern match. If we allow our students to cre-
ate a card of equations and bring it into the exam and then only test the application of equa-
tions, students are likely to forego understanding altogether and only practice equation ma-
nipulation. These kinds of “studying” have minimal impact on real learning and are often
highly ephemeral.
If we really want students to study physics deeply and effectively in a way that will pro-
duce long-term learning, the activities we provide are not enough; the learning they foster
has to be validated via testing in their examinations.
Redish’s sixth teaching commandment: If you want your students to learn something, you
have to test them on it. This is particularly true for items in the “hidden curriculum”
(cf. chapter 3).
Exams as formative feedback

Exams and quizzes are not only means of carrying out summative assessments of what stu-
dents have learned. They can also provide formative feedback to students as to what they have
learned and what they need to work on. Unfortunately, it has been my experience that most
students have the expectation that exams only serve as summative assessments. If they do
poorly, they respond, “Well I sure messed that up. I need to do better on the next topic.” Of
course, since physics is highly cumulative, both in content and in skill building, having a sig-
nificant “hole” in one’s knowledge can lead into a downward spiral.
There are ways of designing the presentation and delivery of exams and quizzes to help
encourage students to pay attention to the mistakes they make on exams. I’ve developed a
pattern of exam delivery that seems to work reasonably well to encourage at least some stu-
dents to debug their thinking from the feedback they get on exams.
• The exam is given in class during the last class of the week.
• The exam is graded immediately (over the weekend) and returned in the first class of the
next week.
4 Be careful! The interpretation that one’s bad responses from students is because one is making them work too hard
is a classic “wishful thinking.”
• I return the exams to the students at the beginning of the class and go over the exam in
class, explicitly showing the partial credit grading patterns, if any.
• I encourage the students to look for grading errors or possible interpretations of the prob-
lem’s wording that would improve their grade. I tell them to submit a written request
for a grade change with a detailed explanation of why they think they should have re-
ceived more points. ( Just saying “please look at problem 4” doesn’t count. They have to
explain why they think they were right and the grader was wrong.)
• If they are dissatisfied with their grade on the exam, they may take an out-of-class makeup
test on the same material (but not the same test). However, the grade on the makeup
does not replace the first test: they get the average of the two grades.
• I tell them that my experience is that a student who got a low grade who simply goes
back and studies again in the same way (or does not study at all) is as likely to go down
on the makeup and lose points as a result. On the other hand, students who specifically
study their first exam in order to understand the mistakes they made and why have a
high probability of raising their grade substantially.
• The class averages on each question are reported and they are told that the final exam
will include at least one question based on the questions the class performed the most
poorly on.
Since this procedure takes up two lecture periods for every exam and requires me to write
two exams for every exam, I tend to reduce the number of hour exams I give in a semester:
two instead of three in a course with three 50-minute lectures per week; one and two shorter
quizzes in a course with two 75-minute lectures per week. In a lecture class of 100 to 200, I
usually get about 25% of the students taking the makeup. This has been effective in making
the students more comfortable with my unfamiliar exam style. (I always include essay ques-
tions, estimation questions, and representation translation questions, along with a more tra-
ditional problem or two.)
The last point in the bulleted list sends the message that I expect them to know every-
thing I am testing them on. If the class as a whole performs badly, I review the physics be-
hind the problem carefully when I go over it in class and tell them that they all need to go
back and look at that material again—and that they are certain to have a question concern-
ing that topic on the final exam.
EIGHT TYPES OF EXAM AND HOMEWORK QUESTIONS

A wide variety of structures are available for presenting physics problems to our students. Dif-
ferent types of structures tend to activate different kinds of associations and resources. A crit-
ical element in developing a course that gets students thinking more deeply about physics is
choosing from a broad palette of question types. In the rest of this chapter, I discuss eight
kinds of problems, briefly describe their value, and give an example or two of each. In addi-
tion to the kinds of problems discussed here, see also the discussion of Peer Instruction and
JiTT in chapter 7 and of Cooperative Problem Solving in chapter 8 where other problem
structures are used.
The Physics Suite has a large collection of problems, being the union of the set devel-
oped over the years for Fundamentals of Physics [HRW6 2001], the set of Workshop Physics
Eight Types of Exam and Homework Questions • 77
problems, and the set I have developed over the years for the Activity-Based Physics project.
These problems are distributed in the text, Understanding Physics, in the problem volume,
and some are available on the Resource CD in the back of this book. Additional sources of
problems and approaches to testing may be found in [Arons 1994], [Tobias 1997], and the
books listed in the file “Other Resources” on the Resource CD associated with this volume.
Multiple-choice and short-answer questions

Multiple-choice and short-answer questions are tempting to use because they are easy to grade.
The results can be highly suggestive, but multiple-choice tests can be difficult to interpret.
They tend to overestimate the student’s learning since they can sometimes be answered cor-
rectly by means of incorrect reasoning5 or by cued responses that fail to represent functional
understanding. On the other hand, the use of common misconceptions or facets as distrac-
tors produces “attractive nuisances” that challenge the students’ understanding. Students who
get the correct answer despite these challenges are likely to have a reasonably good under-
standing. A well-constructed multiple-choice question based on research into common naïve
conceptions can therefore give some indication of the robustness of a student’s correct an-
swer. Note, however, that standard multiple-choice questions developed by instructors who
have not studied the relevant research often have distractors that are too trivial to provide
a real test of students’ understanding. Instructors unaware of research results in physics
education sometimes find it difficult to imagine the kinds of errors that students will com-
monly make.
An example of a good multiple-choice question with research-based distractors is shown in
Figure 4.1. The distractors have to correspond to students’ naïve conceptions, not to what fac-
ulty think. Since most physics instructors know very well that imbedding an object in a fluid
produces a buoyant force, they may find the answers to the problem in Figure 4.1 peculiar. But
A book is at rest on a table top. Which of the following force(s) is (are)

acting on the book?
1. A downward force due to gravity.
2. The upward force by the table.
3. A net downward force due to air pressure.
4. A net upward force due to air pressure.
(A) 1 only
(B) 1 and 2
(C) 1, 2, and 3
(D) 1, 2, and 4
(E) None of these. Since the book is at rest, there are no forces acting on it.
Figure 4.1 A multiple-choice question from the FCI [Hestenes 1992].
5 See, for example, [Sandin 1985].

it is well documented that many high school students and some college students think that air
pressure is responsible for gravity, pushing down on an object with the weight of the air above
it, so item (C) is an attractive distractor for those students.6 A large collection of multiple-choice
questions with research-based distractors is provided in the conceptual surveys of various top-
ics in physics given on the Resource CD associated with this volume.
Short-answer questions can also explore students’ functional understanding of physics
concepts effectively. The key is not to make the question a simple recognition test but to re-
quire some reasoning, perhaps using ideas or principles that are not directly cued by the prob-
lem. In the sample shown in Figure 4.2, the problems naturally cue up buoyant forces, but
they require a fairly sophisticated application of free-body diagrams. (The answer to 2.2, for
example, is obviously “” since both net forces are equal to 0 by Newton’s first law. Students
who do not comfortably distinguish individual and net forces have trouble with this.)
Multiple-choice multiple-response questions

A substantial amount of thinking and reasoning can be required from a student in a multiple-
choice multiple-response test. A sample is shown in Figure 4.3. To solve this problem, a
For each of the following partial sentences, indicate whether they are correctly completed by the
symbol corresponding to the phrase greater than (), less than (), or the same as ().
(2.1) A chunk of iron is sitting on a table. It is then moved from the table into a bucket of water
sitting on the table. The iron now rests on the bottom of the bucket. The force the bucket exerts
on the block when the block is sitting on the bottom of the bucket is _______ the force that the
table exerted on the block when the block was sitting on the table.
(2.2) A chunk of iron is sitting on a table. It is then moved from the table into a bucket of water
sitting on the table. The iron now rests on the bottom of the bucket. The total force on the block
when it is sitting on the bottom of the bucket is _______ it was on the table.
(2.3) A chunk of iron is sitting on a table. It is then covered by a bell jar which has a nozzle con-
nected to a vacuum pump. The air is extracted from the bell jar. The force the table exerts on the
block when the block is sitting in a vacuum is _______ the force that the table exerted on the block
when the block was sitting in the air.
(2.4) A chunk of iron is sitting on a scale. The iron and the scale are then both immersed in a large
vat of water. After being immersed in the water, the scale reading will be _______ the scale read-
ing when they were simply sitting in the air. (Assume the scale would read zero if nothing were sit-
ting on it, even when it is under water.)
Figure 4.2 Sample short-answer question combining the use of different physics principles.
6 Then, Sagredo asks, where does the weight of the air come from? A good question, but one not often asked by
these students.
Four different mice (labeled A, B, C, and D) ran the triangular maze shown below. They
started in the lower left corner and followed the paths of the arrows. The times they took
are shown below each figure.
A B C D
t = 2 sec t = 2 sec t = 4 sec t = 4 sec

For each item below, on your answer sheet write the letters of all of the mice that
fit the description.
(a) This mouse had the greatest average speed.

(b) This mouse had the greatest total displacement.
(c) This mouse had an average velocity that points in this direction (⇒).
(d) This mouse had the greatest average velocity.
Figure 4.3 A sample multiple-choice multiple-response question.
student has to have good control of the concepts of vector displacement and average vector
velocity, and to be able to clearly distinguish velocity and speed.
This type of question requires a student to evaluate each statement and make a decision
about it. It is a particularly useful type of question in cases where students tend to have mixed
context-dependent models of a physical situation.
Representation-translation questions
As discussed in chapter 2, learning to handle the variety of representations we use can be
quite a challenge for introductory students, but it can be one of the most valuable general
skills they develop from studying physics. The presentation of a single situation in different
ways facilitates understanding and sense-making of different facets of the situation. Further-
more, one of the primary difficulties novice students have with problem solving is their fail-
ure to be able to visualize the dynamic situation and map a physics description onto that
visualization.
Thornton and Sokoloff [Thornton 1998] pioneered the use of problems in which stu-
dents are required to match a statement in words describing a physical situation with graphs
or pictures that describe those situations. You will find many such problems in their concept
surveys contained in the Action Research Kit on the Resource CD: the FMCE, the ECCE,
the VET, and the MMCE.
I have had success getting students to think about the meaning of physics variables with
problems in which they are shown a situation and a number of graphs. The graphs have their
abscissa labeled as time, but the ordinates are unmarked. The students then have to match a
list of physical variables to the graphs. An example is shown in Figure 4.4.
I had an interesting experience with a student concerning the representation transla-
tion shown in Figure 4.5. After the exam in which this problem was given, the student (an
An object is clipped to the edge of a disk that is rotating y

with uniform circular motion. At time t = 0 it is at
the position shown and it has the velocity shown. v at t = 0
It travels around with the disk for a full rotation.
A series of graphs is shown below. Identify
x
which of these graphs could represent the following
six items (if it had the appropriate scales):
(a) The x component of the object’s velocity.

(b) The angle the object’s position makes with the x axis.
(c) The y component of the force keeping the object moving in a circle.
(d) The object’s angular velocity.
(e) The object’s speed.
(f) The x component of the object’s position.
A B C
t t
t
D E F
t t
Figure 4.4 A sample representation-translation question.
engineer) came in to complain. “I don’t understand how to do these problems,” he grumped.

I asked what answer he chose for part (a). He reported that he had selected option (e). “And
why did you choose that one?” I asked. “Well,” he said, “it’s a wave and waves are supposed
to be wiggly. That was the wiggliest.”
“Okay,” I responded. “Now tell me what’s happening.”
“What do you mean?” he asked.
I said, “Describe the string, tell me where the dot is. Then tell me what happens to the
string and the dot as the pulse moves down the string.”
“Hmmm,” he said. “Well, the pulse moves to the right, when it gets to the dot, the dot
moves up and then down . . . Oh, [expletive deleted]!” Once he had worked through visu-
alizing what happened, he was almost trivially able to solve the rest of the problem.
Although I have not carried out explicit research on the topic, in my experience, the stu-
dents in both algebra- and calculus-based physics struggle with these problems and the strug-
gle is extremely productive.
Consider the motion of a pulse on a long taut string. We will choose our
coordinate system so that when the string is at rest, the string lies along the x
axis of the coordinate system. We will take the positive direction of the x axis to
be to the right on this page and the positive direction of the y axis to be up.
Ignore gravity. A pulse is started on the string moving to the right. At a time t 0 a
photograph of the string would look like Figure A below. A point on the string to
the right of the pulse is marked by a spot of paint.
For each of the items below, identify which figure below would look most like the
graph of the indicated quantity. (Take the positive axis as up.) If none of the
figures look like you expect the graph to look, write N.
(a) The graph of the y displacement of the spot of paint as a function of time.
(b) The graph of the x velocity of the spot of paint as a function of time.
(c) The graph of the y velocity of the spot of paint as a function of time.
(d) The graph of the y component of the force on the piece of string
marked by the paint as a function of time.
A B C
D E F
Figure 4.5 A sample representation-translation problem.
Ranking tasks
Another class of easy-to-grade but effective problems are ranking tasks —problems in which
the student must order a series. These have been used by a number of researchers and cur-
riculum developers effectively.
Ranking tasks are effective because they easily trigger reasoning primitives7 such as “more
of a cause produces (proportionately) more of an effect” [diSessa 1993]. An example of a
ranking task from the UWPEG is shown in Figure 4.6.
David Maloney has been a long-time user of ranking tasks in his research. Recently, he
and his collaborators, Tom O’Kuma and Curt Heiggelke, published a collection of these prob-
lems [O’Kuma 1999]. A sample is given in Figure 4.7. Many of the Reading Exercises in Un-
derstanding Physics are ranking tasks.
7 Primitives are discussed in detail in chapter 2.

In the picture shown below are 5 blocks. The blocks have equal volumes but
different masses. The blocks are placed in an aquarium tank filled with water,
and blocks 2 and 5 come to rest as shown in the lower figure. Sketch on the
figure where you would expect the blocks 1, 3, and 4 to come to rest. (The
differences in mass between successive blocks is significant—not just a tiny
amount.)
Water level
m1 < m2 < m3 < m4 < m5 2
1 2 3 4 5
5
Figure 4.6 A ranking task [Loverude 1999].
Given below are seven arrangements of two electric charges. In each figure, a
point labeled P is also identified. All of the charges are the same size, Q, but they
can be either positive or negative. The charges and point P all lie on a straight
line. The distances between adjacement items, either between two charges or
between a charge and the point P, are all x cm. There is no charge at point P, nor
are there any other charges in this region.
Rank these arrangements from greatest to least on the basis of the strength of
the electric field at point P. That is, put first the arrangement that produces the
strongest field at point P, and put last the arrangement that produces the
weakest field at point P.
A B C D
P P P P
+ + + − + − − −
E F G
P P P
− + − − + +
Figure 4.7 A ranking task from [O’Kuma 1999].

Context-based reasoning problems

The problems I find most valuable, both on homework and on exams, are those I call
context-based reasoning problems.8 In these problems, students are given a reasonably realistic
situation and have to use physics principles—often in ways or circumstances in that they have
not been previously seen—to come to a conclusion. The crucial fact is that the answer to the
I have set up my two stereo speakers on my back patio as shown in the top view
diagram in the figure at the right. I am worried that at certain positions
I will lose frequencies as a result of interference. The coordinate grid on the
edge of the picture has its large tick marks separated by 1 meter. For ease of
calculation, make the following assumptions:
• Assume that the relevant objects lie on integer or half-integer

grid points of the coordinate system.
• Take the speed of sound to be 343 m/s.
• Ignore the reflection of sound from the house, trees, etc.
• The speakers are in phase.
Speakers
Chairs
Bench
(a) What will happen if I am sitting in the middle of the bench?

(b) If I am sitting in the lawn chair on the left, what will be the lowest frequency I
will lose to destructive interference? (If you do not have a calculator, leave the
result as an expression with numbers that could be simply evaluated and
estimate the result to one significant figure.)
(c) Can I restore the frequency lost in part (a) by switching the leads to one of
the speakers, thereby reversing the phase of that source?
(d) With the leads reversed, what will happen to the sound for a person sitting
at the center of the bench?
Figure 4.8 A context-based reasoning problem.
8 These are similar in spirit to Context-Rich Problems used by the Minnesota group. See the discussion of Coopera-
tive Problem Solving in chapter 8.
problem should be of some reasonable real-world interest. Unfortunately, too many end-of-
chapter physics problems are poorly motivated. They seem simply an exercise in carrying out
some obscure physics calculations for no obvious purpose. A problem such as “How much
work is done by a weight-lifter in raising a 200-kg barbell a distance of 2 meters” is of this
type. Why is the “work” something I should care about? What makes this calculation rele-
vant? A problem such as “Estimate the number of calories a marathon runner burns in
running 26 miles. Does he need to stock up on carbohydrates before beginning?” is better
motivated.
An example that I like is shown in Figure 4.8. This is particularly nice since it relies on
the fundamental idea of interference—figure out the path difference and see how many wave-
lengths fit in. The geometry does not permit the use of the small-angle approximation that
leads to the standard interference formulas. Students have to calculate distances using the
Pythagorean Theorem. It’s not too realistic because with sound, reflections from nearby walls
and objects are of primary importance. But this calculation is one part of understanding what
is going on.
Everyday examples in newspapers, advertising, television, and movies in which the physics
is wrong make nice problems of this type. An example is given in Figure 4.9.
In the movie Jurassic Park, there is

a scene in which some members of
the visiting group are trapped in the
kitchen with dinosaurs on the other
side of the door. The paleontologist
is pressing his shoulder near the cen-
ter of the door, trying to keep out
the dinosaurs. The botanist throws
her back against the door at the edge
next to the hinge. She cannot reach
a gun on the floor because she is try-
ing to help hold the door closed.
Would they improve or worsen their
situation if he moved to the door’s
outer edge and she went for the gun?
Estimate the change in the torque
they would exert on the door if they
changed their positions as suggested.
Figure 4.9 A context-based reasoning problem.

Estimate the number of blades of grass a typical suburban house’s lawn has in the
summer.
Figure 4.10 A typical Fermi question.
Estimation problems
Estimation problems were made famous by Enrico Fermi, who was a master at them. His
classic question was “How many barbers are there in Chicago?” Questions of this type can
have considerable value because students
• get to practice and apply proportional reasoning

• learn to work with large numbers
• learn to think about significant figures (I always deduct points for too many significant
figures.)
• learn to quantify their real-world experience
An example of a typical Fermi question is given in Figure 4.10. I use this question to
explain to my class what it is that I want them to do when they solve problems of this type.9
To get a number, students must picture a lawn, estimate its area, and estimate the number
of blades of grass in a given area of lawn and then scale up. I always require that they start
from a number they can plausibly know. Thus, if they said “let’s assume that a square meter
of lawn has a million blades of grass,” I would give no credit for that estimate. If, on the
other hand, they said,
Consider a square centimeter of grass. There are about 10 blades of grass counting along each
side. So there will be 100 in a square centimeter and 100 100 times that in a square meter.
So assume one million blades of grass per square meter.
that answer would receive full credit. I always grade (or have graded) my estimation prob-
lems so that points are given for each part of the reasoning required. I explain carefully to
my students that they are going to be evaluated on how they come up with the answer, not
just on the answer.
An instructor needs to be both persistent and patient to include estimation problems.
At first, students may not believe that you are serious about asking them to do problems of
this type. For this reason, I give an estimation problem on every homework and exam, and
I identify some homework estimations as having come from previous exams so that students
9 Thisquestion is appropriate for Maryland, where the lawns in the summer still have grass. It might be too easy for
a student at the University of New Mexico in Albuquerque.
will know I’m serious. Second, students may not believe that this is the sort of thing they are
supposed to be learning in physics class and resent doing them at first. One student com-
plained on his post-class anonymous survey that “exam points were given for being a good
guesser.” In my experience, this difficulty passes as they gain skill and confidence. Near the
end of one of my algebra-based physics classes, one of my students told me with great glee,
“Prof, you know those estimation problems? Well I’m taking a business class and we’re do-
ing, like, making a business plan and, you know, it’s just like estimations? I was the only one
in the class who knew how to do it!”
As the class moves on through the year, I begin to blend more physics into estimation
problems so that they become design problems. These seem to play a big role in helping stu-
dents understand the long-term value of physics for their professional future. An example of
an “estimation/design” problem is given in Figure 4.11.
Qualitative questions
Qualitative questions can be quite effective in getting students to learn to think about
concepts—and in helping instructors realize that students may not have learned what they
thought they had from observing good performance on quantitative problem solving.10 The
University of Washington group has been particularly effective in using this kind of question
in curriculum design.11
A sample qualitative problem that is an extension of a ranking task is given in Figure
4.12. I first encountered questions of this type when visiting the University of Washing-
ton on sabbatical in 1992. My inclination was to label each resistor, R, label each battery,
V, write down all the equations using Kirchhoff ’s laws, and solve them for the relevant un-
knowns. This method certainly works! But one of the facilitators12 asked me whether I
could solve it without equations. “Why should I?” I responded. “Because,” he said, “per-
haps your students are not as facile with equations as you are.” I proceeded to give it a try
and was quite surprised at how difficult it was. I realized that I was using the physics equa-
tions as a scaffold for organizing my conceptual knowledge. For students who have not de-
veloped this knowledge structure and who might be hazy about what the concepts mean,
my approach would be largely inaccessible. The approach of reasoning through the prob-
lems conceptually, on the other hand, helps reveal students’ naïve conceptions (such as the
idea that a battery is a source of constant current). As they work through these problems
on homework and struggle with them on exams, they seem to build the concepts more
firmly than they would if they could get away with a purely mathematical approach. Note
that an essential part of questions of this type is the phrase “explain your reasoning.” Many
students don’t know what reasoning or building a case means [Kuhn 1989]. Having them
explain their reasoning, discussing what you mean by reasons, and giving them feedback
on the sort of things we mean by reasoning can be an important part of what they learn
in a physics class.
10 See the discussion of Mazur’s experience in chapter 1.

11 Seethe discussion of Tutorials in Introductory Physics in chapter 8 and Physics by Inquiry in chapter 9.
12 Richard Steinberg, with whom I later had the privilege of collaborating extensively.
You are assigned the task of deciding whether it is possible to purchase a desk-top-
sized magnetic spectrometer in order to measure the ratio of C12 to C14 atoms in a
sample in order to determine its age.
For this probem, let’s concentrate on the magnet that will perform the separation of
masses. Suppose that you have burned and vaporized the sample so that the carbon
atoms are in a gas. You now pass this gas through an “ionizer” that on the average
strips one electron from each atom. You then accelerate the ions by putting them
through an electrostatic accelerator—two capacitor plates with small holes that permit
the ions to enter and leave.
The two plates are charged so that they are at a voltage difference of ∆V Volts. The
electric field produced by charges on the capacitor plates accelerate the ions to an
energy of q∆V. These fast ions are then introduced into a nearly constant, vertical
magnetic field. (See the figure below.) If we ignore gravity, the magnetic field will cause
the charged particles to follow a circular path in a horizontal plane. The radius of the
circle will depend on the atom’s mass. (Assume the whole device will be placed inside a
vacuum chamber.)
Mass spectrometer:
Top view
Collection Large
bins magnet
Accelerator
plates
Magnetic
field
Ion
source
Answer three questions about how the device works.

(a) We would like not to use too high a voltage. If ∆V is 1000 Volts, how big a magnetic
field would we require to have a plausible “table-top-sized” intrument? Is this a
reasonable magnetic field to have with a table-top sized magnet?
(b) Do the C12 and C14 atoms hit the collection plate far enough apart? (If they are not
separated by at least a few millimeters at the end of their path we will have trouble
collecting the atoms in separate bins.)
(c) Can we get away with ignoring gravity? (Hint: Calculate the time it would take the
atom to travel its semicircle and calculate how far it would fall in that time.)
Figure 4.11 An estimation/design problem.

In the circuits below, all the bulbs are identical, and the batteries are identical and
nearly ideal. Rank the brightness of amount of (current through) the bulbs.
Explain your reasoning.
A D E
Figure 4.12 Sample qualitative question on direct current circuits [McDermott 1992].
Of course, in the end I want my students to develop a well-organized knowledge struc-

ture and to have equations mapped firmly into that structure. But it is important not to de-
mand too much too early.
Qualitative questions that require identifying relevant physics principles and concepts,
qualitative reasoning, and writing an explanation can be effective in helping students make
the connection between their real-world personal experiences and the physics they are learn-
ing. An example of a question of this type is given in Figure 4.13.
Essay questions
Essay questions can be the most revealing of students’ difficulties and naïve conceptions of
any form. In the example given in Figure 4.14, the students are not asked to recall the law—
it is given to them. But they are asked to discuss its validity. In this case, we had completed
an ABP Tutorial (see chapter 8) exploring Newton’s third law (N3) with two force probes. I
had wanted them to refer to experimental evidence of some sort in justifying their belief in
N3. Interestingly enough, a significant fraction of students did not expect N3 to hold in all
In public restrooms there are often paper towel dispensers

that require you to pull downward on the towel to extract it.
If your hands are wet and you are pulling with one hand,
the towel often rips. When you pull with both hands, the
towel can be extracted without tearing. Explain why.
Figure 4.13 A sample qualitative reasoning problem that makes a connection to the student’s per-
sonal experience.
Newton’s 3rd law says that objects that touch each other exert forces on each other.
If object A exerts a force on object B, then object B exerts a force back on ob-
ject A and the two forces are equal and opposite.
Consider the following three situations concerning two identical cars and a much
heavier truck.
(a) One car is parked and the other car crashes into it.
(b) One car is parked and the truck crashes into it.
(c) The truck is pushing the car because the car’s engine cannot start. The two
are touching and the truck is speeding up.
In which of these situations do you think Newton’s 3rd law holds or does not hold?
Explain your reasons for saying so. (Your answers are worth 5 points, your rea-
sons 10.)
Figure 4.14 A sample essay question on Newton’s 3rd law.
these cases. The results provided me guidance for follow-up discussions in lecture and later
problems on exams. In the example given in Figure 4.15, I learned in a dramatic fashion that
a large majority of my students had significant difficulties with the concept of “field” despite
my careful attempts to be precise in lecture.
Note that giving essay questions in an exam situation has to be done with some care.
There are three points to consider.
1. In some university environments (such as mine), introductory students often have all
their classes as large lectures and they may have little or no experience with actually
having to write during an examination.
In this semester we have studied two fields: electric and magnetic. Explain why we
introduce the idea of field and compare and contrast the electric and magnetic
fields. In your comparison, be certain to discuss at least one similarity and one
difference.
Figure 4.15 A sample essay question on the field concept.

2. Remember that in exams students are under tremendous emotional pressure. If the
exam is long so that they are also under time pressure, they may not have the op-
portunity to think about what they are writing.
3. The person who learns the most from essay questions on exams is the person grad-
ing them.
As a result of these points, I always try to keep exams short enough that the students are
not overly rushed and I always grade my own essay questions, even if I have a large class and
have TAs available for grading the other parts of the exam.
CHAPTER 5
Evaluating Our
Instruction: Surveys
Mathematics may be compared to a mill of exquisite

workmanship, which grinds you stuff of any degree of fineness;
but, nevertheless, what you get out depends on what you put in;
and as the grandest mill in the world
will not extract wheat flour from peascod,
so pages of formulae will not get a definite result
out of loose data.
T. H. Huxley [Huxley 1869]
As I discussed in the last chapter, there are two ways to probe what is happening in one’s
class. One way is to assess how much each student has learned in order to decide the
extent to which that student receives a public certification of his or her knowledge—a
grade. A second way is to probe our class overall in order to determine whether the
instruction we are delivering is meeting our goals. I refer to the first as assessment, and
the second as evaluation. In the last chapter we discussed how to assess each student so
as to see what he or she was learning. In this chapter, we discuss how to get a snapshot
of how the class is doing overall for the purpose of evaluating our instruction.
As a result of the context dependence of the cognitive response (Principle 2 in chap-
ter 2), in some contexts students may choose to use the model they are being taught,
while in other contexts they may revert to using more naïve resources. When we look
at a class broadly (especially a large class), we can tolerate larger fluctuations in individ-
ual responses than we can when we are assessing individual students. The students’ in-
dividualized choices of what ideas to use and when to use them depend on uncontrol-
lable and unknowable variables (their internal mental states). As a result, their answers
may appear random on a small set of closed-end questions on each topic to be probed.
But these same few questions may give a good average view of what is happening, de-
spite not giving a complete picture of the knowledge of any single student.
91
92 • Chapter 5: Evaluating Our Instruction: Surveys
RESEARCH-BASED SURVEYS
A cost-effective way to determine the approximate state of a class’s knowledge is to use a care-
fully designed research-based survey. By a survey I mean a reasonably short (10- to 30-minute)
machine-gradable test. It could consist of multiple-choice or short-answer questions, or it
could contain statements students are asked to agree or disagree with. It can be delivered on
paper with Scantron™ sheets or on computers.1 It can be delivered to large numbers of stu-
dents and the results manipulated on computers using spreadsheets or more sophisticated sta-
tistical analysis tools.
By research-based, I mean that the survey has been developed from qualitative research
on student difficulties with particular topics and has been refined, tested, and validated by
detailed observations with many students. Broadly, to achieve good surveys (surveys that are
both valid and reliable—see the discussion below) requires the following steps.
• Conduct qualitative research to identify the student models underlying their responses.
• Develop a theoretical framework to model the student responses for that particular topic.
• Develop multiple-choice items to elicit the range of expected possible answers.
• Use the results—including the student selection of wrong answers—to facilitate the de-
sign of new instructions as well as new diagnostic and evaluation tools.
• Use the results to guide construction of new qualitative research to further improve the
survey.
This process places development of evaluational tools firmly in the research-redevelopment cy-
cle of curriculum construction and reform discussed in more detail in chapter 6 (Figure 6.1).
Surveys may focus on a variety of aspects of what students are expected to learn in both
the explicit and hidden curriculum. Content surveys probe student knowledge of the concep-
tual bases of particular content areas of physics. Attitude surveys probe student thinking about
the process and character of learning physics. Over the past two decades, physics education
researchers have developed dozens of surveys that probe topics from mechanics (the Force
Concept Inventory) to the atomic model of matter (the Small Particle Model Assessment
Test). Seventeen such surveys are included on the Resource CD accompanying this volume.
They are listed in the Appendix at the end of this volume.
Why use a research-based survey?

Sagredo scoffs at my emphasis on creating a survey through research. “I’ve given machine-
gradable multiple-choice final exams in my large classes for years. Don’t my grades count as
course evaluations?” Certainly they do, Sagredo. But there are dangers in interpreting exam
results as course evaluations.
The questions we choose often are constrained by a number of factors that may be un-
related to student learning. The first danger is that there is pressure from students (and
sometimes from administrations) to have an “appropriate” grade distribution. Students are
1 Studies to look for differences between paper-delivered and computer-delivered surveys have so far had ambiguous
results.
Research-Based Surveys • 93
comfortable with class averages near 80%, with 90% being the boundary for an A and with
few grades falling below 60%. Although this may make students and administrations happy,
it presses us to produce the requisite number of As no matter what our students have
learned.2
A second danger arises because we are interested in what our students have really learned,
not in what they think you want them to say. By the time they get to university, many stu-
dents have become quite adept at “test-taking skills.” These are even taught in some schools
in order to help students (and school administrators) receive higher evaluations. I’m not crit-
icizing students for taking this approach. I made use of them myself when I took standard-
ized tests. Students taking an exam have the goal to obtain the highest possible score given
what they know. Instructors want the score to accurately reflect what their students know. If
we are not aware in detail of our students’ starting states—what resources and facets are
easily activated—we might be hard pressed to come up with reasonable wrong answers for a
multiple-choice test or with tempting and misleading cues for short-answer questions. With-
out these attractive distractors, students can focus on eliminating obviously wrong answers and
can use their test-taking skills to get a correct result even if they only have a very weak un-
derstanding of the subject.
Neither of these dangers is trivial. The other side of the first danger is that instruction
is never purely objective. It is oriented, in principle, to achieving goals set by the instructor,
though sometimes those goals are tacit, inconsistent, or inappropriate to the particular stu-
dent population involved. An instructor’s exams should reflect his or her own particular learn-
ing goals for his or her students. What is appropriate for us to demand our students learn is
a continuing negotiation among instructors, their students, and outside pressures such as ad-
ministrators, parents, and faculty in the departments our courses serve.
If an instructor is unaware of common student confusions or of how students tend to
respond to particular questions, the result on questions she creates for an exam may not re-
flect what she thinks it does. Furthermore, without a carefully developed question based on
research and a clear understanding of common naïve responses, the students’ wrong answers
may provide little useful information beyond “my students don’t know the answer.”
In trying to interpret the responses of students on closed exam questions, we may en-
counter one or more of the following problems.
1. If a multiple-choice test does not have appropriate distractors, you may not learn what
the students really think.
2. The fact that students give the right answer does not mean they understand why the
answer is right.
3. Since student responses are context dependent, what they say on a single question
only tells part of the story.
4. Problems in the ordering or detailed presentation of the distractors may cause prob-
lems in interpreting the results.
2 In my classes, I try to set exams that have averages between 60% and 65%. A grade over 75% is considered an A.
At this level of difficulty, even the good students get some feedback about where they need to improve. See my
model of examination delivery discussed in chapter 4.
5. It’s easy to overinterpret the implications of a single relatively narrow test—especially

if it only has one style of question.
Concept surveys that are carefully constructed with these points in mind can provide a
useful tool as part of our evaluation of our instruction.
Surveys and the goals of a class

While giving a lecture on physics education research to colleagues in a neighboring physics
department, I once showed some of the questions from the Force Concept Inventory (dis-
cussed in detail below and given on the Resource CD). One faculty member objected quite
vigorously. “These are trick questions,” he said. “What do you mean by a ‘trick question’?” I
asked. He answered, “You really have to have a deep understanding of the underlying physics
to answer them correctly.” After a substantial pause, allowing both him and the rest of the
audience to consider what had just been said, I responded, “Exactly. I want all the questions
I ask to be trick questions.”
This raises a deep question. What is it we want our students to learn from our instruc-
tion? My colleague clearly had much lower expectations for his students than I did—in one
sense. He was satisfied with recognition of an answer but didn’t care if his students could not
distinguish between the correct (physics) answer and an attractive (but incorrect) common-
sense alternative. On the other hand, he probably demands much more in the way of so-
phisticated mathematical manipulations on examinations than I do and is satisfied if his stu-
dents can match a complex problem-solving pattern to one they have memorized. I do not
care if my students can pattern match complex mathematical manipulations. I want them to
be able both to formulate physics problems out of real-world situations and to interpret their
answers sensibly. If we could attain both goals in a one-year course, I would be delighted,
but at present, I don’t know how to do it given the time and resource constraints of a large
class.
This shift in goals can produce some difficulties. Sagredo and I both teach algebra-based
physics on occasion. When he looks at my exams, he complains that they are too easy and
that I’m “dumbing-down” the course. Interestingly enough, many students have reported to
me that the scuttlebutt among the students is “take Sagredo if you want an easy A” and that
my course is the one to take “if you want to work hard and really understand it.” Whenever
Sagredo agrees to give one of my questions to his students on an exam, he is surprised at how
poorly they do. My students would also do poorly on some of his questions.
In the end, when the chalk meets the blackboard, each individual instructor defines his
or her own goals. Nonetheless, there is clearly a need for a community to form to discuss
both the appropriate goals for physics instruction and how to evaluate the extent to which
those goals are reached. That is why I favor the use of research-based surveys as one element
in our evaluations of our instructional success. They are explicit in what they are evaluating,
they are based on careful study of student difficulties, and they are carefully tested for valid-
ity and reliability.
Delivering a survey in your class

Whenever possible, I give pre-post surveys (i.e., at the beginning and end of the class). In
some classes that have an associated lab, the first and last weeks of class do not have labs, and
Research-Based Surveys • 95
so I tell students to come in then to take surveys. In classes that do not have a lab or a blank
week, I am willing to take time in the first and last classes to give surveys. To encourage stu-
dents to take them (and to make them up if they miss the class), I give everyone who com-
pletes each survey 5 grade points (out of a total of about 1000 points for the class as a whole).
If everyone does them, the surveys have no impact on the grading pattern.
To analyze a survey, it is important to compare only matched data sets. That is, only stu-
dents who take both the pre- and the post-tests should be included. This is because there
may be biases in the populations who take the two tests. For example, students who drop the
course in the middle would take the pre-test and not the post-test. If the group of students
dropping the class were biased toward the lower scoring students on the pre-test, this would
bias the pre-post comparison toward high gains. At least if a matched set is used, one is look-
ing at the true gains for a particular set of students.
The danger discussed in the previous section—that students often give us what they
think we want instead of what they think—has three important implications for how surveys
should be delivered, especially if the survey is to meet the purpose of evaluating our instruc-
tion rather than certifying the students. The three implications are:
• We have to be careful to “teach the physics” but not “teach to the test.”
• Survey solutions should not be distributed or “gone over” with the students.
• Surveys should be required (given credit) but not graded.
The first implication, not teaching to the test, is a delicate one. We want the test to
probe students’ knowledge appropriately, and we want our instruction to help them gain the
knowledge that will be probed. Why then is “teaching to the test” usually considered such a
pejorative? I think that it is because in this case we are implicitly using a fairly sophisticated
model of student learning: students should learn how to think, not to parrot back answers
they don’t understand. In our cognitive model (described in chapter 2), this idea can be ex-
pressed more explicitly by saying that students should develop a strong mental model of the
physics with many strong associations that will permit them to identify and use appropriate
solution techniques to solve a wide variety of problems presented in diverse contexts. Re-
search strongly demonstrates that when students learn an answer to a problem narrowly,
through pattern matching, small changes in the problem’s statement can lead to their being
unable to recognize the pattern.3 So if during instruction we give students the specific ques-
tion that will appear on the test, framed exactly as it will be framed there, I call it teaching
to the test.
This leads to the second implication: Survey solutions should not be posted or given out
to students. Research-based survey items can take a long time to develop. Students have to
be interviewed to see how they are reading and interpreting the items and their answers. Sur-
veys have to be delivered repeatedly to study distributions and reliability at a cost to class
time. A carefully developed survey, whatever limitations it may have, is an extremely valuable
resource for the community of physics teachers, but it is fragile. If they are graded and
the answers are posted or discussed in class, they spread—to fraternity/sorority solution
3 I have seen students who solved problems by pattern matching fail to recognize a problem they knew if the pic-
ture specifying the problem was reversed (mirror image).

collections and to student websites—and become a property of the student community rather
than of the instructional community. They are transformed from a moderately effective eval-
uation tool for the teacher to a “test-taking skills” tool for the student.
This leads directly to the third implication: Surveys should not be a part of the student’s
grade. This is a somewhat controversial point. Sagredo suggests that students will not take a
test seriously unless it is graded. This may be true in some populations. At Maryland, it has
been my experience that 95% of my students hand in surveys that show they have been
thought through and answered honestly.4 This might differ in other populations. If a test is
graded, at least some students will make a serious effort to find out the correct answers and
perhaps spread them around. Since I very much don’t want my students to do this, I treat
my exams (which I consider pedagogical tools to help facilitate student learning) and my sur-
veys (which I consider evaluational tools to help me understand my instruction) differently.
There is an additional reason for leaving surveys ungraded. Students often use their test-
taking skills to try to produce an answer that they think the teacher will like, even if they
don’t really think that is the answer. A graded exam definitely tends to cue such responses. I
am more interested in finding out how students respond when such motivation is removed
in order to see whether instruction has had a broader impact on student thinking.
For a survey to be useful, it should be both valid and reliable. I discuss these conditions
next. In the remainder of the chapter I discuss two kinds of surveys that have been devel-
oped and that are currently widely available: content surveys and attitude surveys.
UNDERSTANDING WHAT A SURVEY MEASURES:

VALIDITY AND RELIABILITY
In order to be a useful tool in evaluating instruction, a survey should be valid; that is, it
should measure what it claims to measure. A survey should also be reliable; that is, it should
give reproducible results. When we’re talking about measurements of how people think about
physics instead of about measurements of physical properties, we have to consider carefully
what we mean by these terms.
Validity
Understanding the validity of a survey item, either in a content or attitude survey, is not as
trivial as it may appear on the surface. What’s in question is not just the issue of whether the
physics is right, but whether the question adequately probes the relevant mental structures.
To see what this means, consider the following example. The most common student confu-
sion about velocity graphs is whether they should “look like the motion” or like the rates of
change of the motion. If you ask students in introductory university physics “which graph
corresponds to an object moving away from the origin at a uniform rate” (as in the problem
shown in Figure 5.4) and provide them with the correct (constant) graph but not with the
choice of the attractive distractor (the linearly increasing graph), you will get a high score but
an invalid question. This one is especially subtle. Jeff Saul found that if both of these graphs
4 Evidence to the contrary might be: all answers the same, answers in a recurring pattern, survey completed in one-
fourth the average time, and so on.

Understanding What a Survey Measures: Validity and Reliability • 97
were included but the correct answer given first, 80% of the students selected the right an-
swer. But if the attractive distractor (a rising straight line) was given first, the success rate fell
by about a factor of 2 [Saul 1996].
In order to achieve validity, we need to understand quite a bit not only about the physics
but also about how students think about the physics. Since human beings show great flexi-
bility in their responses to situations, we might expect an intractably large range of responses.
Fortunately, in most cases studied, if the topic is reasonably narrowly defined, a fairly small
number of distinct responses (two to ten) accounts for a large fraction of the answers and
reasonings that are found in a given population. Understanding this range of plausible vari-
ation is absolutely essential in creating valid survey items. As a result, the first step in creat-
ing a survey is to study the literature on student thinking and do extensive qualitative re-
search to learn how students think about the subject.
Even if we understand the range of student thinking on a topic, for an item to be valid
students must respond to it in the expected way. This is one of the most frustrating steps in
creating a survey. A good way to probe this is to observe a large number of students “think-
ing aloud” while doing the survey. Culture shifts and vocabulary differences between the (usu-
ally middle-aged) test designers and (usually young adult) subjects can produce dramatic sur-
prises. In our validation interviews for the Maryland Physics Expectations (MPEX) survey
discussed later in this chapter, we wanted to know if students understood our motivation for
doing derivations in lecture. To our dismay we learned that a significant fraction of our cal-
culus-based introductory physics students were unfamiliar with the word “derivation” and
thought it meant “derivative.”5 To get a consistent valid response we had to rephrase our
items.
Reliability
Reliability also has to be considered carefully. In probing human behavior, this term replaces
the more standard physics term repeatability. We usually say that repeatibility means that if
some other scientists repeated our experiment, they would get the same result, within ex-
pected statistical variations. What we really mean is that if we prepare a new experiment in
the same way as the first experimenter did, using equivalent (but not the same) materials, we
would get the same result. We don’t expect to be able to measure the deformability of a piece
of metal many times using the same piece of metal. We are comfortable with the idea that
“all muons are identical,” so that repeating a measurement of muon decay rates doesn’t mean
reassembling the previously measured muons out of their component parts.
But when it comes to people, we are accustomed to the idea that people are individuals
and are not equivalent. If we try to repeat a survey with a given student a few days later, we
are unlikely to get the identical result. First, the “state of the student” has changed somewhat
as a result of taking the first survey. Second, the context dependence of the cognitive response
reminds us that “the context” includes the entire state of the student’s mind—something over
which we have little control. Experiences between the two tests and local situations (Did a bad
exam in another class make her disgruntled about science in general? Did an argument with
5 This is perhaps a result of the unfortunate strong shift away from “proof ” in high school math classes that took
place in the 1990s.

his girl friend last night shift his concerns and associations?) may affect student responses. And
sometimes, students do, in fact, learn something from thinking about and doing a survey.
Fortunately, these kinds of fluctuations in individual responses tend to average out over
a large enough class. Measures can become repeatable (reliable) when considered as a mea-
sure of a population rather than as a measure of an individual. However, we must keep in
mind that according to the individuality principle (Principle 4, chapter 2), we can expect a
population to contain substantial spreads on a variety of measures. Any survey is measuring
a small slice of these variables. There is likely to be a significant spread in results, and the
spread is an important part of the data.6
In the educational world, reliability testing is sometimes interpreted to mean that stu-
dents should respond similarly to the same question formulated in different ways. For ex-
ample, one might present an “agree–disagree” item in both positive and negative senses. For
the MPEX, discussed below, we have the following pair of items to decide whether a student
feels that she needs to use her ordinary-world experiences in her physics class.
To understand physics, I sometimes think about my personal experiences and relate them to
the topic being analyzed.
Physics is related to the real world and it sometimes helps to think about the connection, but
it is rarely essential for what I have to do in this course.
Although these items are designed to be the reverse of each other, they are not identical. The
difference between “sometimes” and “rarely essential” can lead a student to split the differ-
ence and agree with both items, especially if that student is on the fence or is in transition.
Even when items are closer than these, students can hold contradictory views. In this case,
the “lack of reliability” in the responses to the matched questions lies not in the test but in
the student. Care must be taken not to eliminate questions that show this kind of “unrelia-
bility,” lest one bias the survey toward only seeing topics on which most students have formed
coherent mental models.
CONTENT SURVEYS
In the remainder of this section I discuss three of the most commonly used surveys in me-
chanics in detail: the Force Concept Inventory (FCI), the Force and Motion Conceptual Eval-
uation (FMCE), and the Mechanics Baseline Test (MBT). The additional content surveys
that are included on the CD provided with this book are listed and described briefly in the
Appendix.
The FCI
One of the most carefully researched and most extensively used concept surveys in our cur-
rent collection is the Force Concept Inventory (FCI) developed by David Hestenes and his
collaborators at Arizona State University [Hestenes 1992a]. This is a 30-item multiple-choice
6 A useful metaphor for me is a spectral line. For many physical circumstances, the width and shape of the line is
important data, not just its centroid.

Content Surveys • 99
survey meant to probe student conceptual learning in Newtonian dynamics. It focuses on is-
sues of force (though there are a few kinematics questions), and it is easily deliverable. Stu-
dents typically take 15 to 30 minutes to complete it.
Building a probe of student conceptual understanding requires both understanding the
fundamental issues underlying the physics to be probed, as viewed by the professional sci-
entist, and understanding the common naïve conceptions and confusions the students spon-
taneously bring to the subject as a result of their experience. In creating their mechanics
concept test, Hestenes and his collaborators first thought carefully about the conceptual
structure of Newtonian mechanics. But understanding the professional’s view is not enough.
The test has to be designed to respond properly when considered from the student’s point
of view. The distractors (wrong answers) should distract! That is, there should be answers
that correspond to what many naïve students would say if the question were open ended
and no answers were given.
Hestenes and his collaborators relied on existing research and did extensive research of
their own to determine spontaneous student responses [Halloun 1985a] [Halloun 1985b]
[Hestenes 1992a]. They then compiled a list of common naïve conceptions and attempted
to create questions that would reveal whether or not the students harbored these naïve con-
ceptions. Their list of naïve conceptions is given in the FCI paper [Hestenes 1992a]. Many
of them are directly related to the facets created applying primitive reasoning in the context
of motion. (See chapter 2.)
Finally, in constructing the FCI, Hestenes and his collaborators chose to use semirealis-
tic situations and everyday speech rather than technical physics speech in order to set the con-
text to be the student’s personal resources for how the world works rather than what one is
supposed to say in a physics class. See, for example, the upper part of Figure 2.6, and Figure
5.1. Answer (C) in Figure 5.1 is an example of a research-based distractor. Few physics in-
structors who have not studied the research literature would think of choosing such an item;
Imagine a head-on collision between a large truck and a small compact car. Dur-
ing the collision:
(A) the truck exerts a greater amount of force on the car than the car exerts on
the truck.
(B) the car exerts a greater amount of force on the truck than the truck exerts
on the car.
(C) neither exerts a force on the other; the car gets smashed simply because it
gets in the way of the truck.
(D) the truck exerts a force on the car but the car does not exert a force on the
truck.
(E) the truck exerts the same amount of force on the car as the car exerts on
the truck.
Figure 5.1 A question from the Force Concept Inventory [Hestenes 1992a].
80
1.00
HS COLL UNIV
% Gain = (% Post-test − % Pre-test)

Interactive Engagement
Traditional
60 0.69
40
0.34
20
0.23
slope = Gain/(Max. Possible Gain)

0
20 40 60 80 100
% Pre-test
Figure 5.2 A plot of class average pre-test and post-test FCI scores for a collection of classes in high
school, college, and university physics classes using a variety of instructional methods [Hake 1992].
it’s not even “on their screen” as a possible wrong answer. But a significant number of naïve
students, unaccustomed to seeking forces from objects as the cause of all changes in motion,
actually select this answer.7
The FCI is perhaps the most widely used concept survey in the nation today. Its publi-
cation in 1992 stirred a great deal of interest in physics education reform among the com-
munity of college and university physics instructors. Looking at the test, most faculty de-
clared it “trivial” and were shocked when their students performed poorly.8 A typical score
for a class of entering calculus-based physics students is 40% to 50% and a typical score for
a class of entering algebra-based physics students is 30% to 45%. At the completion of one
semester of mechanics, average scores tend to rise to about 60% for calculus-based students
and 50% for algebra-based students. These are rather modest and disappointing gains.
Richard Hake of Indiana University put out a call for anyone who had given the FCI
pre-post in a college class to send him their results, together with a description of their class.
He collected results from over 60 classes [Hake 1992]. His results are displayed in Figure 5.2
and show an interesting uniformity. When the class’s gain on the FCI (post-test average–
pre-test average) is plotted against the class’s pre-test score, classes of similar structure lie ap-
proximately along a straight line passing through the point (100,0). Traditional classes lie on
the line closest to the horizontal axis and show limited improvement. The region between
7 College students who have previously taken high school physics are less likely to choose (C) as an alternative here.
They are more likely to select an “active agent” primitive or a facet built on a “more is more” primitive and to se-
lect (A).
8 Compare the Mazur story in chapter 1. Mazur was influenced by the earlier version of the Hestenes test [Halloun
1985a], as was I.
the two dotted lines represents classes with more self-reported “active engagement.” Hake
claims that classes lying near the line falling most steeply reported that they were using
active-engagement environments and a research-based text. This suggests that the negative
slope of the line from a data point to the point (100,0) is a useful figure of merit:
g (class post-test average class pre-test average)/(100 class pre-test average)
where the class averages are given in percents.

The interpretation of this is that two classes having the same figure of merit, g, have
achieved the same fraction of the possible gain—a kind of educational efficiency. Hake’s results
suggest that this figure of merit is a way of comparing the instructional success of classes with
differently prepared populations—say, a class at a highly selective university with entering
scores of 75% and a class at an open enrollment college with entering scores of 30%. This
conjecture has been widely accepted by the physics education research community.9
Hake’s approach, though valuable as a first look, leaves some questions unanswered. Did
people fairly and accurately represent the character of their own classes? Did a selection oc-
cur because people with good results submitted their data while people with poor results chose
not to? To answer some of these questions, Jeff Saul and I undertook an investigation of 35
classes at seven different colleges and universities [Redish 1997] [Saul 1997]. Four different
curricula were being used: traditional, two modest active engagement reforms (Tutorials and
Group Problem Solving: one hour of reform class per week), and a deeply reformed high
active-engagement curriculum (Workshop Physics) in early implementations.10 We gave pre-
post FCI in each class and observed the classes directly. The FCI results are summarized in
Figure 5.3.
These results confirm Hake’s observations and give support to the idea that g is one plau-
sible measure of overall gain. Some additional interesting conjectures may be made after study-
ing this figure.
1. In the traditional (low-interaction lecture-based) environment, what the lecturer does

can have a big impact on the class’s conceptual gains.
The peak corresponding to the traditional class is very broad. At Maryland, where we ob-
served the largest number of classes in a reasonably uniform environment, the classes with
the largest gains were taught by award-winning professors who tried to actively engage their
classes during lecture. The lowest gains were taught by professors who had little interest in
qualitative or conceptual learning and focused their attention on complex problem solving.
(For the detailed “unsmoothed” version of the Maryland results, see Figure 8.3.)
2. In the moderate active-engagement classes (one modified small-class group-learning

hour per week), much of the conceptual learning relevant to FCI gains was occur-
ring in the modified class.
9 Some detailed preliminary probes of this question have been reported at meetings of the American Association of
Physics Teachers (AAPT), but no decisive publications have yet resulted.
10 See chapters 8 and 9 for detailed descriptions of the methods.
Traditional
Tutorial/Cooperative
Problem Solving
Workshop Physics
(Early adopters)
0 0.1 0.2 0.3 0.4 0.5 0.6

h
Figure 5.3 A plot of the fractional FCI gain achieved in three types of classes: traditional, moderate
active engagement (tutorial/group problem solving), and strong active engagement (early adopters of
workshop physics). Histograms are constructed for each group and fit with a Gaussian, which is then
normalized [Saul 1997].
This is suggested by the narrowness of the peak and the fact that it lies above the results at-
tained by even the best of the instructors in the traditional environments.
3. Full active-engagement classes can produce substantially better FCI gains, even in
early implementations.
This is suggested by the results from the Workshop Physics classes studied. For a more de-
tailed discussion of this issue (and for the results from mature Workshop Physics at the pri-
mary site), see chapter 9.
Although the FCI has been of great value in “raising the consciousness” of the commu-
nity of physics teachers to issues of student learning, it has its limitations. Besides those lim-
itations associated with all machine-gradable instruments, it is lacking in depth on kinemat-
ics issues that are to some extent a prerequisite to understanding the issues probed. A more
comprehensive survey is provided by the Force and Motion Conceptual Evaluation.
The FMCE
The Force and Motion Conceptual Evaluation (FMCE) was developed by Ron Thornton and
David Sokoloff [Thornton 1998]. In addition to the dynamical issues stressed by the FCI,
this survey addresses student difficulties with kinematics, especially difficulties with repre-
sentation translation between words and graphs. It is longer than the FCI, with 47 items in
a multiple-choice multiple-response format that is somewhat more difficult for students to
untangle than the (mostly) straightforward FCI multiple-choice items. As a result, students
need more time to complete the FMCE—from 30 minutes to an hour.
An example of an FMCE item is given in Figure 5.4. Although this looks superficially
trivial, students have a great deal of difficulty in choosing the correct graphs until they have
Questions 40–43 refer to a toy car which can move to the right or left along a horizontal line
(the positive portion of the distance axis). The positive direction is to the right.
0 +
Choose the correct velocity-time graph (A - G) for each of the following questions. You may
use a graph more than once or not at all. If you think that none is correct, answer choice J.
A + E +
Velocity
Velocity
0 0
0 Time 0 Time
− −
B + F +
Velocity
Velocity
0 0
0 Time 0 Time
− −
C + G +
Velocity
Velocity
0 0
0 Time 0 Time
− −
D + H +
Velocity
Velocity
0 0
0 Time 0 Time
− −
J None of these graphs is correct.
40. Which velocity graph shows the car moving toward the right (away from the origin) at
a steady (constant) velocity?
41. Which velocity graph shows the car reversing direction?
42. Which velocity graph shows the car moving toward the left (toward the origin)
at a steady (constant) velocity?
43. Which velocity graph shows the car increasing its speed at a steady (constant)
rate?
Figure 5.4 An MCMR set of items from the FMCE [Thornton 1998].
clearly differentiated the concepts of velocity and acceleration and have developed good graph-
mapping skills.11 (In my own use of this survey, I exchange graphs (A) and (D) so as to bet-
ter probe how many students are tempted to assign the “linearly rising” graph to a constant
velocity.)
The FMCE is structured into clusters of questions associated with a particular situation,
as shown in Figure 5.4. This tends to “lock” students into a particular mode of thinking for
the cluster of problems and may not give a clear picture of the range of student confusion
on a particular topic [Bao 1999].
11 By “graph-mapping” skills, I mean the ability to map a physical situation onto a variety of different graphical
representations.
1.00 1.00
0.80 0.80
FCI fractional score
0.60 0.60
0.40 0.40
0.20 0.20
Correlation Coefficient = 0.791 Correlation Coefficient = 0.8309
0.00 0.00
0.00 0.20 0.40 0.60 0.80 1.00 0.00 0.20 0.40 0.60 0.80 1.00
FMCE fractional score FMCE fractional score
Figure 5.5 Scatter plot of FMCE versus FCI scores pre (left) and post (right). The size of the mark-
ers indicates the number of students with those scores [Thornton 2003].
Some of the items in the FMCE are included to “set up” the student’s frame of mind or
to check for students who are not taking the test seriously (e.g., item 33). All students are
expected to get these correct since they cue widely held facets that lead to the correct result.
See the description of the FMCE analysis in the file on the CD. (The FMCE analysis tem-
plate included on the CD is already set up to handle this.)
Thornton and his collaborators have carried out extensive studies of the correlation be-
tween FMCE results and FCI results [Thornton 2003]. They find a very strong correlation
(R 0.8) between the results, but the FMCE appears to be more challenging to low-scoring
students, with few students scoring below 25% in the FCI while FMCE scores go down to
almost 0%. Scatter plots of pre- and post-FCI versus FMCE scores are shown in Figure 5.5.
(The areas of the circles are proportional to the number of students at the point.)
The MBT
Both the FCI and the FMCE focus on components of basic conceptual understanding and
representation translation. Our goals for physics classes at the college level usually include
applying conceptual ideas to solve problems. Hestenes and his collaborators created the
Mechanics Baseline Test (MBT) to try to probe students’ skill at making these connections
[Hestenes 1992b]. Scores tend to be lower than on the FCI. Although this survey is designed
for the introductory physics class, David Hestenes told me that when he gave it to the physics
graduate students in his first-year graduate classical mechanics class, it correlated well with
their grades. An example of an MBT item is given in Figure 5.6. In this item, students have
to recognize the relevance of energy conservation. Students who fail to do so tend to activate
various facets or other associations.
Attitude Surveys • 105
10. A young girl wishes to select one of the frictionless playground slides illustrated
below to give her the greatest possible speed when she reaches the bottom of the slide.
2.5 m
A B C D
0.5 m
Which of the slides illustrated in the diagram above should she choose?
(A) A (B) B (C) C (D) D

(E) It doesn’t matter; her speed would be the same for each.
Figure 5.6 An item from the MBT [Hestenes 1992b].
ATTITUDE SURVEYS
If we want to understand whether our students are making any progress on our hidden cur-
riculum of learning both process and scientific thinking, we need to find some way to probe
the state of their attitudes.12 One approach that has provided a useful first look is to use an
attitude survey. Three attitude surveys are provided on the CD accompanying this book: the
MPEX, the VASS, and the EBAPS.
In using an attitude survey, one needs to be aware of some limitations and caveats. First,
attitudes, like most thinking process, are complex and context dependent. But they may fluc-
tuate more widely than narrower content knowledge topics. The attitudes toward learning
that students bring to our classroom may vary from day-to-day, depending on everything
from whether they attended a party instead of studying the night before to whether a pro-
fessor in another class has given a difficult and time-consuming homework assignment. Sec-
ond, students’ understanding of their own functional attitudes may be limited. Surveys of at-
titudes only measure what students think they think. To see how they really think, we have
to observe them in action.
The MPEX
We created the Maryland Physics Expectations (MPEX) survey in the mid-1990s to provide
a survey that could give some measure of what was happening to our students along the di-
mensions of relevance to the hidden curriculum [Redish 1998]. The focus of the survey was
not on students’ attitudes in general, such as their epistemologies or beliefs about the nature
of science and scientific knowledge, but rather on their expectations. By expectations we mean
that we want the students to ask themselves: “What do I expect to have to do in order to
12 See chapter 3 for more discussion of the hidden curriculum.

TABLE 5.1 The Items of the MPEX Reality Cluster.
#10: Physical laws have little relation to what I experience in the real world. (D)
#18: To understand physics, I sometimes think about my personal experiences and relate them to the
topic being analyzed. (A)
#22: Physics is related to the real world and it sometimes helps to think about the connection, but it
is rarely essential for what I have to do in this course. (D)
#25: Learning physics helps me understand situations in my everyday life. (A)
succeed in this class?” I emphasize the narrowness of this goal: “this class,” not “all my sci-
ence classes” or “school in general.”
The MPEX consists of 34 statements with which the students are asked to agree or dis-
agree on a 5-point scale, from strongly agree to strongly disagree.13 The MPEX items were
validated through approximately 100 hours of interviews, listening to students talk about each
item, how they interpreted it, and why they chose the answer they did. In addition, the par-
ity of the favorable MPEX responses was validated by offering it to a series of expert physics
instructors and asking what answers they would want their students to give on each item [Re-
dish 1998]. The desired parity (agree or disagree) is labeled the favorable response, and the
undesired parity is labeled unfavorable.
To illustrate the MPEX focus on expectations, consider the items given in Table 5.1. The
favorable response (agree A or disagree D) is indicated at the end of the item. These
items ask students to evaluate the link between physics and their everyday experience in two
ways: from the class to their outside experience and from their outside experience to the class.
Each direction is represented by two elements: one to which the favorable response is posi-
tive and one to which it is negative. The more general item “Physics has to do with what
happens in the real world” was omitted, since almost all students agreed with it.
MPEX results
We analyze and display the results on the MPEX by creating an agree/disagree (A/D) plot. (See
Figure 5.7.) In this plot, “agree” and “strongly agree” are merged (into “A”), and “disagree”
and “strongly disagree” are merged (into “D”). The result is a three-point scale: agree, neu-
tral, and disagree. This collapse of scale is based on the idea that, while it might be difficult
to compare one student’s “strongly agree” to another student’s “agree,” or to make much of
a shift from “strongly agree” to “agree” in a single student, there is a robust difference be-
tween “agree” and “disagree” and a shift from one to the other is significant. The unfavorable
responses are plotted on the abscissa and the favorable responses on the ordinate. Since the
A D N (“neutral”) responses must add to 100%, the point representing a class lies within
the triangle bounded by the abscissa (unfavorable axis), the ordinate (favorable axis), and the
line representing 0% neutral choices (F U 100). The expert responses are plotted as a
cross in the favorable corner of the triangle.
13 In education circles, such a ranking is referred to as a Likert (Lı̆k-ert) scale.

Experts
Favorable
UMN
OSU
UMD
0 10 20 30 40 50 60 70 80 90 100
Unfavorable
Figure 5.7 An agree/disagree (A/D) plot of overall MPEX results in large university settings at the
beginning and end of a one-semester calculus-based physics class. All classes had traditional lectures
with a one-hour active-engagement small-class session per week. Each represents data from 500 stu-
dents [Redish 1998].
Pre- and post-scores can be plotted for each cluster on the A-D plot. It is convenient to
connect the pre- and post-results for each cluster with an arrow. A spreadsheet that allows
you to paste in your MPEX results and generate A-D plots is included on the Resource CD
associated with the volume.14 Overall data from three large state universities are shown in
Figure 5.7 (data from [Redish 1998]).
The MPEX has been delivered as a pre-post survey to thousands of students around the
United States. The results have been remarkably consistent.
1. On the average, college and university students entering calculus-based physics classes
choose favorable response on approximately 65% of the MPEX items.
2. At the end of one semester of instruction in large lecture classes, the number of fa-
vorable responses drops by approximately 1.5. This is true even in classes that con-
tain active-engagement elements that produce significantly improved conceptual gains
as measured, say, by the FCI.
14 This spreadsheet was created by Jeff Saul and Michael Wittmann.

Analyzing the MPEX

Among the 34 items of the MPEX, 21 are associated into five clusters corresponding to the
three Hammer variables described in chapter 3, plus two more. The five MPEX clusters are
described in Table 5.2.
Note that some of the MPEX items belong to more than one cluster. This is because the
MPEX variables are not interpreted as being linearly independent. The breakdown into clus-
ters of the MPEX results at Maryland in the first semester of engineering physics is shown
in Figure 5.8. These results are averaged over seven instructors and represent a total of 445
students (matched data). We see that there are significant losses in all of the clusters except
concepts where we had a small gain. Note that not all of the MPEX items have been assigned
to clusters.
Items 3, 6, 7, 24, and 31 (an “Effort cluster”) are included to help an instructor under-
stand what the students actually do in the class. They include items such as
#7: I read the text in detail and work through many of the examples given there. (A)
#31: I use the mistakes I make on homework and on exam problems as clues to what I need
to do to understand the material better. (A)
Although students’ answers to these items are interesting, I recommend that they not be in-
cluded in overall pre-post analyses. There is a strong tendency for students to hope that they
will do these things before a class begins, but they report that they didn’t actually do them
after the class is over. Inclusion of these items biases the overall results toward the negative.
MPEX items 1, 5, 9, 11, 28, 30, 32, 33, and 34 are not assigned to clusters. Interviews
suggest that these items are indeed correlated with student sophistication, but they do not
correlate nicely into clusters. Furthermore, since the MPEX was designed for a class in cal-
culus-based physics for engineers, some of these items may not be considered as desirable
goals for other classes.
TABLE 5.2 The MPEX Variables and the Assignment of Elements to Clusters.
Favorable Unfavorable MPEX Items
Independence Takes responsibility for constructing Takes what is given by authorities 8, 13, 14, 17, 27
own understanding (teacher, text) without evaluation
Coherence Believes physics needs to be Believes physics can be treated as 12, 15, 16, 21, 29
considered as a connected, unrelated facts or independent
consistent framework “pieces”
Concepts Stresses understanding of the Focuses on memorizing and using 4, 14, 19, 23, 26, 27
underlying ideas and concepts formulas without interpretation
or “sense-making”
Reality Believes ideas learned in physics Believes ideas learned in physics 10, 18, 22, 25
are relevant and useful in a wide are unrelated to experiences
variety of real contexts outside the classroom
Math link Considers mathematics as a Views the physics and math 2, 8, 15, 16, 17, 20
convenient way of representing independently with little
physical phenomena relationship between them
100
90
80
70
Math
60
Favorable
Reality
Independence
50 Coherence
40 Concepts
30
20
10
0 20 40 60 80 100
Unfavorable
Figure 5.8 Pre-post shifts on the MPEX clusters at the University of Maryland in the first semester
of engineering physics (data from [Redish 1998]).
Two items in particular tend to be controversial.
#1: All I need to do to understand most of the basic ideas in this course is just read the text,
work most of the problems, and/or pay close attention in class. (D)
#34: Learning physics requires that I substantially rethink, restructure, and reorganize the
information that I am given in class and/or in the text. (A)
Sagredo is unhappy about these. He says, “For #1, I would be happy if they did that.
Why do you want them to disagree? For #34, some of my best students don’t have to do this
to do very well in my class. Why should they agree?” You are correct, Sagredo, and I suggest
that you give these items, but not include them in your analysis or plots.15 We include them
because our interviews have revealed that the best and most sophisticated students in a class
who are working deeply with the physics respond favorably as indicated. Certainly, for in-
troductory courses this level of sophistication may be unnecessary. I like to retain these items,
hoping that something I am doing is helping my students realize that developing a deep un-
derstanding of physics requires the responses as indicated.
Getting improvements on the MPEX

The fact that most courses probed with the MPEX show losses is discouraging but not un-
expected. It is not surprising that students do not learn elements of the hidden curriculum
15 This is easily achieved in the Saul-Wittmann MPEX analysis spreadsheet by replacing “1”s by “0”s in the appro-
priate cells.
90%
Math
80%
Reality Coherence
70% Overall
Concepts
60% Independence
50%
40%
30%
20%
10%
10% 20% 30% 40% 50% 60% 70% 80% 90%
Figure 5.9 An A/D plot for pre-post MPEX results for Andy Elby’s physics class at Thomas Jefferson
High School in Virginia. For each cluster, the pre-result is at the base of the arrow, the post is at the
tip of the arrow, and the name of the associated cluster is next to the arrowhead. The overall result is
shown as a gray arrow [Elby 2001].
as long as it stays hidden. If we want students to improve along these dimensions, we have
to be more explicit in providing structures to help them learn them.
Recently, MPEX results in classes designed to focus on explicit instruction in intuition
building, coherence, and self-awareness of one’s physics thinking have shown substantial im-
provements in all the MPEX categories [Elby 2001]. These results are shown in Figure 5.9.
I have been able to achieve MPEX gains in my large lecture classes by making specific
efforts to keep issues of process explicit in both lectures and homework. Soon after develop-
ing the MPEX in 1995, I made strong efforts in my calculus-based physics classes to produce
gains by giving estimation problems to encourage a reality link, by talking about process, and
by stressing interpretation of equations in lecture. Results were disappointing. The responses
on the reality link items still deteriorated, as did overall results. After much thought and ef-
fort, I introduced activities in lecture to help students become more engaged in these issues
(see the discussion of Interactive Lecture Demonstrations in chapter 8), and I expanded my
homework problems to include context-related problems every week. I reduced the number
of equations I used and stressed derivations and the complex application of the few concep-
tually oriented equations that remained. The results (in my algebra-based class in 2000) were
the first MPEX gains I had ever been able to realize.16 Some of the results for four interest-
ing items are shown in Table 5.3.
16 This class also produced the largest FCI/FMCE gains I have ever managed to achieve.
TABLE 5.3 Pre- and Post-results on Four MPEX Items from a Calculus-Based Class Using
UW Tutorials and Algebra-Based Class Using More Explicit Self-Analysis Techniques.
Calculus-based Algebra-based
(1995) (2000)
F U N F U N
#4 “Problem solving” in physics Pre 60% 21% 19% 66% 30% 4%
basically means matching
problems with facts or Post 77% 13% 10% 91% 9% 0%
equations and then
substituting values to
get a number.
#13 My grade in this course is Pre 54% 24% 22% 57% 40% 3%
primarily determined by
how familiar I am with Post 49% 23% 28% 79% 19% 2%
the material. Insight or
creativity has little to
do with it.
#14 Learning physics is a matter Pre 39% 28% 33% 36% 53% 11%
of acquiring knowledge
that is specifically located Post 37% 24% 39% 64% 34% 2%
in the laws, principles,
and equations given in
class and/or in the
textbook.
(#19) The most crucial thing in Pre 43% 32% 25% 45% 45% 10%
solving a physics problem
is finding the right Post 46% 26% 28% 72% 26% 2%
equation to use.
The MPEX serves as a sort of “canary in the mine” to detect classes that may be toxic
to our hidden curriculum goals. The fact that most first-semester physics classes result in a
deterioration of favorable results is telling. The fact that gains can be obtained by strong and
carefully thought out efforts suggests that the use of the MPEX can be instructive, when ju-
diciously applied.
The VASS
A second survey on student attitudes toward science was developed by Ibrahim Halloun and
David Hestenes [Halloun 1996]. The Views about Science Survey (VASS) comes in four
forms: one each for physics, chemistry, biology, and mathematics. The physics survey has 30
items. Each item offers two responses, and students respond to each item on an eight-point
scale as shown in Figure 5.10. (Option 8 is rarely chosen.) In addition to items that probe
what I have called expectations, the survey includes items that attempt to probe a student’s
epistemological stance toward science. A sample item is given in Figure 5.11.
1 2 3 4 5 6 7 8
Equally Neither
Towards “Only (a)” Towards “Only (b)”
(a) & (b) (a) nor (b)
Figure 5.10 The eight-point scale for responding to items from the VASS [Halloun 1996].
The VASS is designed to probe student characteristics on six attitudinal dimensions—

three scientific and three cognitive.
Scientific dimensions of the VASS

1. Structure of scientific knowledge: Science is a coherent body of knowledge about pat-
terns in nature revealed by careful investigation rather than a loose collection of di-
rectly perceived facts (comparable to MPEX coherence cluster).
2. Methodology of science: The methods of science are systematic and generic rather than
idiosyncratic and situation specific; mathematical modeling for problem solving in-
volves more than selecting mathematical formulas for number crunching (extends
MPEX math cluster).
3. Approximate validity of scientific results: Scientific knowledge is approximate, tentative,
and refutable rather than exact, absolute, and final (not covered in the MPEX).
Cognitive dimensions of the VASS

4. Learnability: Science is learnable by anyone willing to make the effort, not just by a
few talented people, and achievement depends more on personal effort than on the
influence of teacher or textbook.
5. Reflective thinking: For a meaningful understanding of science one needs to concen-
trate on principles rather than just collect facts, look at things in a variety of ways,
and analyze and refine one’s own thinking.
6. Personal relevance: Science is relevant to everyone’s life; it is not of exclusive concern
to scientists (relates in part to MPEX reality cluster).
The favorable polarization of the VASS responses was determined by having it filled out
by physics teachers and professors. Teachers’ responses were strongly polarized on most items.
The laws of physics are:

(a) inherent in the nature of things and independent of how humans think.
(b) invented by physicists to organize their knowledge about the natural world.
Figure 5.11 A sample item from the VASS [Halloun 1996].

TABLE 5.4 Categories for Classifying VASS Responses [Halloun 1996].

Profile Type Number of Items out of 30
Expert 19 or more items with expert views
High Transitional 15–18 items with expert views
Low Transitional 11–14 items with expert views and an equal or small number
with folk views
Folk 11–14 items with expert views but a larger number of items
with folk views or 10 items or less with expert views
Answers agreeing with the teachers are called expert views, while the polar opposites are re-
ferred to as folk views. Halloun and Hestenes classify students into four categories depend-
ing how well their responses agree with those of experts (see Table 5.4).
Halloun and Hestenes delivered the VASS to over 1500 high school physics students in 39
schools (30 of which used traditional rather than active engagement methods) at the beginning
of class. They found them to be classified about 10% expert, about 25% high transitional, about
35% low transitional, and about 30% folk. Surveys of beginning college physics students gave
similar results. For the high school students, there was a significant correlation between the stu-
dents’ profiles on the VASS and their gains on the FCI, as shown in Figure 5.12.
70
60
Percentage of high school students
Profiles
EP
50 HTP
LTP
FP
40
30 Post-test% − Pre-test%
g=
100 − Pre-test%
20
10
0
g ≤ .23 .23 < g < .52 g ≥ .52
low gain moderate gain high gain
Gain factor
Figure 5.12 Correlation between VASS profiles and student gains on the FCI [Halloun 1996].
The EBAPS
Both the MPEX and the VASS suffer from the problem of probing what students think they
think rather than how they function. In addition, they have the problem that for many
items, the “answer the teacher wants” is reasonably clear, and students might choose those
answers even if that’s not what they believe. In the Epistemological Beliefs Assessment for
Physics Science (EBAPS), Elby, Fredericksen, and White attempt to overcome these prob-
lems by presenting a mix of formats, including Likert-scale items, multiple-choice items,
and “debate” items. Many EBAPS items attempt to provide context-based questions that ask
students what they would do rather than what they think. The debate items are particularly
interesting. Here’s one.
#26:
Justin: When I’m learning science concepts for a test, I like to put things in my own
words, so that they make sense to me.
Dave: But putting things in your own words doesn’t help you learn. The textbook was
written by people who know science really well. You should learn things the way the
textbook presents them.
(a) I agree almost entirely with Justin.
(b) Although I agree more with Justin, I think Dave makes some good points.
(c) I agree (or disagree) equally with Justin and Dave.
(d) Although I agree more with Dave, I think Justin makes some good points.
(e) I agree almost entirely with Dave.
The EBAPS contains 17 agree-disagree items on a five-point scale, six multiple-choice items,
and seven debate items for a total of 30. The Resource CD includes the EBAPS, a descrip-
tion of the motivations behind the EBAPS, and an Excel template for analyzing the results
along five axes:
Axis 1 Structure of knowledge

Axis 2 Nature of learning
Axis 3 Real-life applicability
Axis 4 Evolving knowledge
Axis 5 Source of ability to learn
CHAPTER 6
Instructional Implications:
Some Effective Teaching Methods
In theory, there is no difference

between theory and practice.
But in practice, there is.
Jan L. A. van de Snepscheut
quoted in [Fripp 2000]
The information in previous chapters describing what we know about the general char-
acter of learning and about the general skills we are trying to help students develop has
profound implications for building an effective instructional environment. Sagredo once
asked me, “OK. You’ve told me all this stuff about how people learn and shown me lots
of references about specific student difficulties with particular bits of physics content.
Now tell me the best way to teach my physics class next term.”
Sorry, Sagredo. I wish it were that straightforward. First, as I pointed out in chapter
2, no single approach works for all students. Both individual differences and the particu-
lar populations in a class need to be taken into account. Second, despite the great progress
in understanding physics learning that has been made in the past two decades, we’re still
a long way from being able to be prescriptive about teaching. All we can give are some
guidelines and a framework for thinking about what might work for you. Third, the de-
cisions teachers (or a department) make about instruction depend very strongly on the
particular goals they would like to achieve with a particular course. Traditionally, these
goals have been dominated by surface features rather than by deep structure—by se-
lecting specific content matched, perhaps, to the long-term needs of the population be-
ing addressed rather than by thinking about student learning and understanding. The ed-
ucation research described above allows us to expand our community’s discussion about
what different students might learn from taking a particular physics course. This discus-
sion has only just begun, and it is really only in the context of such a discussion that spe-
cific optimized curricula can be developed. Our goal is to transform good teaching from
115
116 • Chapter 6: Instructional Implications: Some Effective Teaching Methods
an art that only a few can carry out to a science that many can learn, but we’ve not got-
ten that far yet.
The traditional approach to physics at the college level involves lectures with little
student interaction, end-of-chapter problem solving, and cookbook labs. Although students
who are self-motivated independent learners with strong mathematical and experimental
skills thrive in this environment (as they do in almost any educational environment), this
category represents only a small fraction of our students. Indeed, the group seems to be
shrinking, since young people today rarely have the “hands-on” mechanical experience
common to physicists “of a certain age” and their teachers. The self-motivated indepen-
dent learners of today are much more likely to have created their own computer games
than to have built a crystal radio, rebuilt the engine of their parents’ Ford, or been in-
spired by Euclid’s Elements.
At present, we not only know a lot about where and why students run into diffi-
culties, but the community of physics educators has developed many learning environ-
ments that have proven effective for achieving specific goals. With the Physics Suite, we
pull together and integrate a number of these environments. In this chapter, I give brief
overviews of innovative curricular materials that have been developed in conjunction with
careful research, including both Suite elements and other materials that work well with
Suite elements.
Before discussing specific curricula, however, I briefly discuss what I mean by a “re-
search-based curriculum,” describe the populations for which these curricula have been
developed, and consider some of the specific goals that are being addressed. After this
preamble, I briefly list the curricular materials of the Physics Suite and a few others that
have been developed that match well with the Suite. In the next three chapters, I discuss
these materials in detail.
RESEARCH-BASED CURRICULA
Most of the curricula that have been developed over the past few years in the United States
are based at least in part on a model of student thinking and learning1 similar to the one de-
scribed in chapter 2 and have evolved using the cyclic model of curriculum development that
I refer to as the research-redesign wheel. In this process, shown schematically in Figure 6.1,
research on student understanding illuminates the difficulties in current instruction. The re-
sults of the research can be used to design new curricula and teaching approaches that lead
to modified instruction. Research and evaluation informs on the state of effectiveness of the
instruction and illuminates difficulties that remain. This process begins again and cycles in a
helix of continuous educational improvement.
1 In some, the dependence on a model of thinking and learning is tacit.

Research-Based Curricula • 117
RESEARCH/
EVALUATION
CURRICULUM
Model
of the DEVELOPMENT
Learner
INSTRUCTION
Figure 6.1 The Research and Redesign Wheel—the role of research in curriculum reform.
Of course, to understand what one sees in a research situation, one must have a model
or theory of the system under investigation in order to know what to look for and to make
sense of what one sees. On the other side, the experimental observations may cause us to re-
fine or modify our theoretical model. So to the wheel, I add an axle—with the model of cog-
nition and learning serving as the point about which the wheel rotates.
The research and evaluation components in this model lead to a cumulative improve-
ment of the curriculum that is usually absent when individual faculty members develop ma-
terials in response to local needs. In the three decades that I have been a faculty member at
the University of Maryland, I have watched my colleagues, highly intelligent, dedicated to
their educational tasks, and concerned about the students’ lack of learning in the laboratory,
modifying and redesigning the laboratories for populations of students ranging from pre-
service teachers to engineers and physics majors. Each faculty member changes something he
or she finds ineffective and makes what he or she thinks is an improvement. But since the
purpose of the change is not shared, since the value of the change is not documented, and
since the culture of instruction tends to focus on the individual instructor’s perception of
what is good instruction, the next instructor is likely to undo whatever changes have been
made and make new changes. Instead of a cumulative improvement, the curriculum under-
goes a drunkard’s-walk oscillation.2 The addition of the research/evaluation component to the
cycle and the input from our theoretical understandings of the student and of the learning
process enable us to produce curricula that can be considerably more effective than those pro-
duced by individual faculty working alone.
2 Perhaps one can expect a long-term improvement but only proportional to the square root of the time!
MODELS OF THE CLASSROOM

Most physics instruction in the United States is delivered in one of two kinds of environ-
ments: the traditional, instructor-centered structure, and an active-engagement student-cen-
tered structure.
The traditional instructor-centered environment

If the class is large (50), there are usually three hours of class per week, with all the stu-
dents meeting together. Often, there is a weekly two- or three-hour laboratory associated with
the class, uncoordinated with the lecture. If there is sufficient staff (such as graduate students
to serve as teaching assistants), there may be one or two hours a week of recitation—a ses-
sion in which the class is divided into small groups (30). This traditional model of intro-
ductory physics has a number of characteristics. As taught in the United States it has the fol-
lowing common features:
• It is content oriented.
• If there is a laboratory, it is two to three hours and “cookbook” in nature; that is, stu-
dents will go through a prescribed series of steps in order to demonstrate the truth of
something taught in lecture or read in the book.
• The instructor is active during the class session, while the students are passive (at least
during lectures and often during recitation).
• The instructor expects the students to undergo active learning activities on their own
outside of the class section, in reading, problem solving, etc., but little or no feedback
or guidance is given to help the students with these activities.
For most students, the focus of the class is the lecture. The nature of this experience can
be seen clearly in the structure of the classroom. A typical lecture room is illustrated in Fig-
ure 6.2. All students are turned to face the lecturer—the focus of all attention. There may be
a strong tendency for the instructor to do all the talking and to discourage (or even to sup-
press) student questions or comments.
The active-engagement student-centered environment

An active-engagement class has somewhat different characteristics.
• The course is student centered. What the students are actually doing in class is the focus
of the course.
• Laboratories in this model are of the guided discovery type; that is, students are guided
to observe phenomena and build for themselves the fundamental ideas via observation.
• The course may include explicit training of reasoning.
• Students are expected to be intellectually active during the class.
Active-engagement classes may occur as part of a larger class—as a recitation or labora-

tory combined with a traditional lecture. The smaller units have a classroom structure that
Models of the Classroom • 119
Figure 6.2 A typical lecture classroom. Even when the lecturer is superb, the focus of the activity
tends to be on the lecturer, not the students. (Here, Jim Gates presents one of his popular public lec-
tures on string theory. Courtesy Dept. of Physics, Univ. of Maryland.)
looks something like Figure 6.3. Students’ attention is focused on their work and on their
interaction with the other students in their group. Facilitators roam the room while the stu-
dents are working, checking the students’ progress and asking guiding questions. There may
be one or more facilitators, and they may be faculty, graduate assistants, undergraduates who
have had the class previously, or volunteers looking to gain teaching experience.
I refer to such an arrangement as an active-engagement classroom. Of course, the struc-
ture of the room does not guarantee what will happen in that room. You can do a mindless
Figure 6.3 The arrangement of an active-engagement classroom [Steinberg 2001].

cookbook lab in one of these classrooms just as easily as a highly effective discovery lab. But
the structure of the room does constrain the possibilities. You can do activities in this kind
of room that would be extremely difficult to carry out in a large lecture hall.
A specific type of active-engagement classroom is the workshop or studio class. In this en-
vironment, the lecture, laboratory, and recitation are combined in a single classroom. In work-
shop classes, most of the class time is taken up by periods in which the students are actively
engaged in exploring the physics using some laboratory equipment, often involving comput-
ers in order to allow efficient high-quality data collection and to provide computer model-
ing tools. Only a small fraction of the period may be spent with a teacher lecturing to the
students. One example of a workshop classroom is the interesting layout developed for Work-
shop Physics by Priscilla Laws and her collaborators at Dickinson College (see Figure 6.4).
Students work two per computer station at tables with two stations. The tables are shaped so
that neighboring pairs can easily collaborate. The room is set up so that there is a group in-
teraction space in the center where demonstrations can be carried out and where the teacher
can stand and easily view what is on every computer screen. This feature has the great ad-
vantage of helping the instructor identify students who might be in trouble or not on task.
There is a table with a screen and blackboard at one end so that the instructor can model
problem solving, do derivations, or display simulations or videos. The materials developed
for Workshop Physics are a part of the Physics Suite and are discussed in detail in chapter 9.
Figure 6.4 A typical workshop or studio classroom layout (Courtesy Kerry Browne, Dickinson College).
The Population Considered: Calculus-Based Physics • 121
Other arrangements for workshop-style classes have been developed at RPI for Studio
Physics and at North Carolina State for the SCALE-UP project. The SCALE-UP project is
discussed as a case study for the adoption and adaptation of Suite materials in chapter 10.
There is evidence that active-engagement characteristics alone do not suffice to produce
significant gains in student learning [Cummings 1999]. The presentation of traditional ma-
terials in an active-engagement learning environment does not necessarily result in better con-
cept learning than a traditional environment. What seems to be necessary is that specific at-
tention is paid to the knowledge and beliefs students bring into the class from their experience and
previous instruction.
THE POPULATION CONSIDERED: CALCULUS-BASED PHYSICS

As of this writing, the process of research-based curriculum development is farthest along for
the introductory calculus-based (“university”) physics course and the course taken by pre-
service elementary school teachers. The Physics Suite primarily addresses the former group
(though Physics by Inquiry and Explorations in Physics specifically address the latter).
Characteristics of calculus-based physics students

Calculus-based (university) and algebra-based (college) physics courses are usually the largest
service courses presently offered by physics departments.3 At present, most of the curricular
materials that have been developed have been created with the calculus-based physics class in
mind. The students in this class have a number of characteristics that distinguish them from
other students.
• They are mostly mathematically relatively sophisticated.

• Almost all have studied physics in high school and done well in it.
• Almost all think physics is important for their careers.
• They mostly consider themselves scientists or engineers.
The hidden curriculum and problem solving

When we talk about our classes, we usually specify a certain set of content. But if all our stu-
dents take away from our course is content, their ability to use this content may be limited.
They may have developed a vocabulary, learned to recognize that they’ve seen a particular
equation before, and may perhaps have improved their algebraic skills somewhat. This is not
enough to keep students taking these courses at a time when there is great competition for
places in the engineering curriculum, and it does not scratch the surface of the powerful and
valuable skills and attitudes that could be delivered. I refer to the (usually tacit) gains that we
hope our students will achieve as a result of taking a physics course as the hidden curriculum.
3 There are some tantalizing counterexamples that illustrate possibilities for future developments. One example is
Lou Bloomfield’s class for nonscientists at the University of Virginia, “How Things Work,” using his text of the same
name [Bloomfield 2001]. As of this writing, I am unaware of any research on the results of this class on student
learning or understanding.
We began to discuss the hidden curriculum in chapter 3. Here, let’s try to explicate some of
those elements that might be important for developing authentic problem-solving skills, based
on the understanding of student learning we have developed in previous chapters.
The research on problem solving shows that experts use a good understanding of the
concepts involved to decide what physics to use. Novices look for an equation. Experts clas-
sify problems by what physics principles are most relevant, such as energy vs. force analysis.
Novices classify them by surface structure and superficial associations (e.g., it’s an inclined
plane), and they remember a particular problem they did with inclined planes [Chi 1981].
We would really like our students to learn the components of problem solving used by ex-
pert physicists:
• The ability to “find what physics will be useful” for a problem

• The skill to take apart and solve complex problems
• The ability to evaluate the result of a solution and know whether it makes sense
In order to achieve all of these goals, a student has to be able to make sense of what
a problem “is about.” In order to develop such a mental model, an understanding of the
concepts—of the physical meaning of the terms and symbols used in physics—is essential
(necessary, but not sufficient). As described in chapter 1, success in algorithmic problem solv-
ing has been shown to be poorly correlated with a good understanding of basic concepts
[Mazur 1997] [McDermott 1999]. This observation fits well with the cognitive structures de-
scribed in chapters 2 and 3.
SOME ACTIVE-ENGAGEMENT STUDENT-CENTERED CURRICULA

The physical (and temporal) architecture of the classroom is only one part of what controls
what happens to students; the other is the cognitive architecture, which is determined by the
curricular materials and by how the instructor uses them. The curricula associated with the
Physics Suite, and the additional curricula I have chosen to include, are coherent, rely on the
educational principles discussed in the first half of this book, and focus on getting students
“to do what needs to be done.” Most of them focus on the goal of improving student con-
ceptual understanding and their ability to use these concepts in complex problem solving.
Models of instruction have been developed that replace one or more of the elements of
the traditional structure by an active-engagement activity. Lecture-based models modify the tra-
ditional lecturer presentation to include some explicit student interaction. Laboratory-based
models replace the traditional laboratory by a discovery-type laboratory. Recitation-based mod-
els replace the recitation in which an instructor models problem solving for an hour by a
structure in which the students learn reasoning or problem solving in groups guided by work-
sheets. They may also carry out qualitative guided-discovery experiments. Finally, some mod-
els go beyond the traditional structure by creating an environment that combines elements
of lecture, laboratory, and recitation in a single class, usually dominated by guided-discovery
laboratories. I refer to these as workshop models.
Following are the models that I discuss in the next few chapters. The specific materials
that have explicitly been coordinated as part of the Physics Suite are marked in bold.
Some Active-Engagement Student-Centered Curricula • 123
Lecture-based models (chapter 7)

• Traditional lecture
• Peer Instruction/ConcepTests
• Interactive Lecture Demonstrations
• Just-in Time Teaching
Recitation-based models (chapter 8)
• Traditional recitation
• Tutorials in Introductory Physics
• ABP Tutorials
• Cooperative Problem Solving
Laboratory-based models (chapter 8)
• Traditional laboratory
• RealTime Physics4
Workshop models (chapter 9)
• Physics by Inquiry
• Workshop Physics
• Explorations in Physics (not discussed in this volume)
In the next three chapters, the discussion of each model begins with a boxed summary;
each summary describes briefly the following elements:
• The environment in which the method is carried out (lecture, lab, recitation, or work-
shop)
• The staff required to implement the method
• The populations for which the method has been developed and tested and those to
whom it might be appropriately extended
• Whether computers are required to implement the method and how many
• Other specialized equipment that might be required
• The time investment needed to prepare and implement the method
• The materials and support that are available
Within the description of the method itself, I discuss the method briefly, consider some
explicit example, and, if there is data on the method’s effectiveness, I present some sample
data. If I have had personal experience with the method, I discuss it.
4Tools for Scientific Thinking, a somewhat lower level set of laboratory materials similar in spirit to RealTime Physics,
are also a part of the Physics Suite but are not discussed in this volume.
CHAPTER 7
Lecture-Based Methods
When I, sitting, heard the astronomer,

where he lectured with such applause in the lecture room,
How soon, unaccountable, I became tired and sick;
Till rising and gliding out, I wander’d off by myself,
In the mystical moist night-air, and from time to time,
Look’d up in perfect silence at the stars.
Walt Whitman
Most of the introductory physics classes in the United States rely heavily on the tradi-
tional lecture. Research has rather broadly shown (see, e.g. [Thornton 1990]) that lec-
tures, even when given by good lecturers, have limited success in helping students make
sense of the physics they are learning. Good lectures can certainly help motivate stu-
dents, though, as I discuss in chapter 3, lecturers often don’t know how to help students
convert that motivation to solid learning.
Even in a traditional class with a large number of students, there are some things
you can do to get your students more engaged during a lecture. Unfortunately, some of
the “obvious” things that both Sagredo and I have tried to do in lecture—such as ask-
ing rhetorical questions, asking them to think about something I’ve said, telling them to
make a prediction before a demonstration (but not making those predictions public), hav-
ing them work out something in their notebooks, or even doing lots of demonstrations—
don’t seem to have much effect. Something more structured seems to be required—
something that involves having them give explicit responses that are collected and paid
attention to.
In this chapter, I discuss my experience in traditional lectures and provide some de-
tailed tips that in my experience can help improve that environment. Then I describe
three models that involve more structured interactions with the students and that have
been shown to produce dramatic improvements in student learning: Peer Instruction,
Interactive Lecture Demonstrations, and Just-in-Time Teaching.
124
The Traditional Lecture • 125
THE TRADITIONAL LECTURE
Environment: Lecture.
Staff: One lecturer (N 20–600).
Population: Introductory algebra- or calculus-based physics students.
Computers: None required.
Other Equipment: Traditional demonstration equipment.
Time Investment: 10–20 hours planning time per semester, 1–2 hours preparation
time per lecture.
Traditional lectures offer opportunities to inspire and motivate students, but one shouldn’t
make the mistake of assuming that students immediately understand and learn whatever the
professor says and puts on the board. I consider myself a good “learner-from-lectures,” hav-
ing had years of experience with them. I regularly attended my lectures in university and grad
school, took excellent notes, and studied from them. As a researcher, first in nuclear physics
and now in physics education, I attend dozens of seminars and conference lectures every year.
I enjoy them and feel that I learn from them.
But occasionally, I’ve been brought up short and have been reminded what the experi-
ence is like for a student. I still vividly recall a colloquium given some years ago at the Uni-
versity of Maryland by Nobel Laureate C. N. (Frank) Yang on the subject of magnetic
monopoles. This was a subject I had looked at briefly while in graduate school, had some in-
terest in, but hadn’t pursued at a professional level. I had all the prerequisites (though some
were rusty) and was familiar with the issues. Yang gave a beautiful lecture—clear, concise,
and to the point. I listened with great pleasure, feeling that I finally understood what the is-
sues about magnetic monopoles were and how they were resolved. Leaving the lecture, I ran
into one of my friends and colleagues walking toward me. “Oh, Sagredo,” I said. “You just
missed the greatest talk!”
“What did you learn?” he asked.
I stopped, thought, and tried to bring back what I had just heard and seen in the lec-
ture. All I could recall was the emotional sense of clarity and understanding—but none of
the specifics. I was left with the only possible response, “Frank Yang really understands
monopoles.”
Once I had my grad assistants (ones not associated with the course) stationed waiting at
the top of my lecture hall after lecture, grabbing students leaving at the end of the class and
asking them “What did he talk about today?” Of the students willing to stop and chat, al-
most none could recall anything about the lecture other than the general topic.
This could in principle be an acceptable situation. If I had taken good lecture notes in
Yang’s lecture, I could have gone back to look at them and spent the time weaving the new
information into my existing schemas. (I thought I had “listened for understanding” instead.)
Unfortunately, many of my students do not take good lecture notes. Of those who do, many
do not know how to use them as a study aid. Of those who know how to use their notes,
many are highly pressed for time by other classes, social activities, or jobs, and can’t devote
126 • Chapter 7: Lecture-Based Methods
the time required for the task. The assumption that “most students use lectures to create good
lecture notes which they then study from” can be a very bad assumption indeed.
A more interactive approach to the traditional lecture

An alternative approach is to use the lecture in a more interactive way. Even within the frame-
work of the traditional lecture, there are many tricks the instructor may use to increase the
student’s intellectual engagement in the class.1 Some are fairly obvious and are taught in classes
for new faculty, probed in end-of-the-semester student questionnaires, and watched for by
peer evaluators. They include:
• Speak clearly and at an appropriate pace.

• Write on the board using good handwriting and good layout.
• Give students sufficient time to copy anything you expect them to copy.
Lecturers are often not aware of defects along these lines, focusing on the content rather than
on what they are saying about the content. Problems with these issues can be helped by video-
taping and reviewing your lectures or by having a sympathetic colleague sit in the back and
watch your presentation.
Some of the things you can do to keep the students interested and attentive are the
following:
• Set the context.

• Chunk the material.
• Facilitate note-taking.
• Develop a good speaking technique.
• Ask authentic questions.
• In discussions, value process as well as right answers.
• Get students to vote on a choice of answers.
• Make it personal.
Set the context

I once heard Sagredo deliver a lecture to a graduate class of physics students on a subject I
thought I needed to learn more about. The lecture was enlightening—and I learned what I
needed to know—but he presented it in a way I found disturbing. He began with 45 min-
utes of technical development without any discussion of his motivation or why this devel-
opment was going to be interesting or useful. In the last 5 minutes, he wrapped everything
together in an elegant package, applying all the technical details to the case of interest. When
I asked him why he approached the lecture in this way, he said, “I didn’t want to give away
the punch line.” Sagredo, I think stand-up comedy is the wrong metaphor for a physics
1 Many of these and more are discussed in Donald Bligh’s useful book, What’s the Use of Lectures? [Bligh 1998].
lecture. Although our students seem to be accustomed to trying to take in random, unmoti-
vated mathematical results, I don’t think that is the best way to engage their attention and
interest.
Since everyone’s thinking and learning is naturally associative, we can expect to get the
best results by tying new material to something the student already knows.2 I try to begin
my lectures by setting a context, letting the students know beforehand what the point of the
lecture is and where we are going. Before every lecture, I write an outline of what we will be
doing on the upper-left corner of the board so that students can have some idea of what we
are going to be talking about.
Chunk the material

Another thing I try to do when lecturing in a large hall with many students is to keep in
mind the difficulty produced by the limits to working memory. You cannot expect your stu-
dents to keep a large number of difficult ideas in mind for a long time and bring them to-
gether at the end as you tie everything up into a neat package.
I try to chunk my lectures into coherent pieces that begin at the upper left and that can
be completed on the available board space. Once I’ve finished the chunk, I don’t just con-
tinue, but I stop, walk to the back of the class, wait until the students finish their note-
taking, and go over the entire argument again so that the students have a context and can
see the entire presentation at once. Summarizing after the chunk is complete helps students
find a way to integrate the new material with their existing knowledge structures.
Facilitate note-taking
Tricks to speed things up, such as using pre-prepared transparencies, are usually counterpro-
ductive, especially if you are expecting students to take notes. Copying something from the
board usually takes more time than it does to write it on the board, since the copyist also has
to read and interpret what’s been written. Forcing yourself to write it on the board at least
gives you some idea of what the students are going through. Even if you do not expect the
students to copy from a display, you are likely to severely underestimate the time it takes the
students to read and make sense of the material you put up since you are very familiar with
it and they are not.
Handing out previously prepared lecture notes may in principle help a bit, but since the
instructor (rather than the student) is preparing the notes, the instructor gains the associated
processing benefits, not the student. A more plausible approach3 is to provide students with
a set of “skeleton” notes—with just the main points sketched out but with spaces in which
the students are supposed to fill-in what happens in class. This could be useful in helping
students to create a well-organized set of notes and might help them in following the lecture.
I have adapted this idea by creating PowerPoint presentations for each lecture that have a sim-
ilar “skeleton” structure. The students can choose to print these out as handouts. I do deri-
vations and problem solutions on the board so they don’t go by too quickly—but the figures
and diagrams can be more neatly prepared on the computer.
2 Recall the example with the strings of numbers in chapter 2.

3I learned this trick from Robert Brown at Case-Western Reserve University.
Develop a good speaking technique

Getting students in a large class to engage the material is most easily accomplished by hav-
ing them engage in carefully designed individual and group activities. The research-based cur-
ricular materials discussed below give some examples of how this can be done. Even without
the use of prearranged materials, you can improve your students’ engagement somewhat in
these ways:
• Talk to the students —Face the students when you speak. If you have to write something
on the board, do not “talk to the board” with your back to the students. Write and then
turn to the class to describe or explain it.
• Use appropriate tones of voice—Learn to project your voice. Test your classroom with a
friend, seeing if you can be heard adequately from all parts of the room. If you can’t, use
a microphone. Be careful! It’s natural to project in a loud voice during the test and to
forget to do it when you get involved in what you are saying in lecture. One trick that
seems effective is, after announcing that something is important, drop your voice a bit
to present the important information. The class will quiet significantly to hear what you
are saying.
• Step out of the frame—In a lecture hall, walk up the aisles and speak from the middle or
the back of the class. This requires the students to turn around. Changing their orien-
tation restores their attention (at least momentarily—and it allows you to stand next to
and stare down a student who is reading a newspaper). This also breaks the imaginary
“pane of glass” the students put up between you and them and helps to change you from
a “talking head on a TV screen” (to whom they feel no need to be polite or considerate)
to a human being (to whom they do).
• Make eye contact —When you look a student in the eye during your lecture, for that stu-
dent, you change the character of the activity from a TV- or movie-like experience into
one more like a conversation, even if it’s only for a moment, and even if you are doing
most of (or all of ) the talking. But be careful not to fixate on one particular student.
That can be intimidating for that student. Switch your gaze from student to student
every few seconds.
Ask authentic questions

An excellent way to get students involved is to ask questions to which they respond by really
thinking about and answering them. This can be very effective, but it is harder than it sounds.
Most faculty questions are rhetorical—that is, they are not meant to be answered by the
students—in practice, if not in intent. Faculty tend to be as nervous about “dead air” as a TV
news anchor. Two or three seconds of silence can seem like an eternity while you are waiting
for students to answer a question you’ve posed. The easiest solution is to answer it yourself.
But students know that faculty do this, so they wait you out. To get them to realize that you
really do want them to answer and that the question is not just a part of the lecture, you have
to outwait them—at least until they get in the habit of answering. This can be quite painful
until you get used to it. The idea is to wait until they get uncomfortable with the silence, and
this can easily take 20 to 30 seconds or longer. Sometimes you may have to reiterate your ques-
tion or call on a specific student at random to show you really want a response.
You also have to build your students’ confidence that answering questions is not going
to be a painful experience. Students are most reluctant to look foolish in front of their in-
structor (and in front of their peers), so it is often quite difficult to elicit responses to ques-
tions. Being negative or putting down a student’s question or answer can in one sentence es-
tablish a lack of trust between the instructor and the class that can last for the rest of the
term. The result can be a class in which the instructor does all the talking, severely reducing
the students’ involvement and attention. I feel strongly enough about this principle that I set
it off as
Redish’s seventh teaching commandment: Never, ever put down a student’s comment in
class or embarrass a student in front of classmates.
This isn’t always easy, even for a supportive and compassionate lecturer. I remember one oc-
casion some years ago in which I was lecturing to a class of about 20 sophomore physics ma-
jors on the Bohr model. As I proceeded to put down a blizzard of equations (laid out most
clearly and coherently, I was certain), one student stopped me and asked: “Professor Redish,
how did you get from line 3 to line 4?” I looked at the equations carefully and mentally re-
minded myself that the student asking the question was mathematically quite sophisticated
and was taking our math course in complex variables that semester. After a pause during which
I suppressed a number of put-downs and nasty remarks, I responded: “You multiply both sides
in line 3 by a factor of two,” before proceeding without further comment. Thinking about the
situation later, I realized that I had been going too fast, using too many equations without suf-
ficient explanation, and not giving the students enough time to follow the argument.
In discussions, value process as well as right answers

Another important step in being able to build a class that responds and participates in dis-
cussion is to change the class’s idea that you are looking for the “right answer.” If that’s all
you ever ask for, only the few brightest and most aggressive students will answer your ques-
tions. This will only reconfirm the attitude that most students have that science in general
and physics in particular is a collection of facts to be memorized rather than a process of
reasoning—and that only a few really bright people can do it. (See chapter 3.)
Even if the first student answering your question gives the correct answer, one way to
begin to break this epistemological misconception is to ask for other possible answers, em-
phasizing creativity and explaining that the students are not required to believe the answers
they give. I’ll give an example of my experience with this technique in the section under In-
teractive Lecture Demonstrations.
Get students to vote on a choice of answers

Even if only a small number of students are willing to respond to an instructor’s question,
there are still ways of engaging a larger fraction of students in a lecture. One of the easiest
and most effective is voting. This is easy to implement. Set out some options and ask the
class to raise their hands in support of the different options. In some classes, only a few stu-
dents will respond. If the voting process is to work to keep them engaged, this has to be over-
come. In such a case, I often walk up the aisle and point to someone who hasn’t voted and
ask them to explain their difficulty and why they were unable to make a decision.
Another idea is to give each student a set of five “flash cards” at the beginning of the
term.4 These should be large (the size of a notebook page) and have one of the letters “A”
through “E” on one side, and other options (“true,” “false,” “yes,” “no,” and “maybe” or “?”)
on the back. It may help to make them different colors. The students are instructed to bring
their flashcards to every lecture. (If you really want them to do this, you have to use them at
least once in every lecture, preferably more often than that.) When you present your choices,
you label them with the letters or other options, and the students hold up the answer they
choose. You can easily see the distribution of answers. If the flashcards are not colored, it’s
important to report back to the students an approximate distribution of the voting. If you
use colored cards, they will be able to see each other’s cards. The sense of not being alone in
their opinions is an important part of increasing their comfort with taking a stand that might
turn out to be wrong. There are electronic versions of this system available in which each stu-
dent gets a remote-control device on which they can click their answers. These answers are
beamed to a collector and displayed on a computer projection screen.
Make it personal
Finally, perhaps the most important component of delivering effective lectures (and classes in
general) is to show the students that you are on their side.
Redish’s eighth teaching commandment: Convince your students that you care about their
learning and believe that they all can learn what you have to teach.
This can make a tremendous difference in a class’s attitude, no matter what environment you
are using. One way to demonstrate this caring is to learn as many of the students’ names as
you can. Even if you only learn the names of the students who ask questions in class, it will
give the rest of the students the impression that you know all (or most) of them. I take pho-
tographs in recitation section and copy them. (This also helps my teaching assistants learn
the students’ names.) I then bring them with me to class and spend three to five minutes be-
fore class matching names to faces. After class, I check any students who have come up to
ask questions. It turns out to be relatively easy to learn the names of 50 to 100 students with-
out much effort. Not only do the students get more personally engaged in the class, but
so do I.
Demonstrations
An important component of a traditional introductory physics lecture is the lecture demon-
stration. Sagredo suggested to me that perhaps he should just do more demonstrations “es-
pecially since they don’t seem to follow the math very well. After all, seeing is believing.” I
wish it were that easy, Sagredo. As physicists, we are particularly enamored of a good demon-
stration. After all, if they are properly set up, a demonstration makes clear what the physics
is—doesn’t it?
4I learned this idea from Tom Moore. See [Moore 1998] and [Meltzer 1996].
Unfortunately, demonstrations are not always as effective as we expect them to be, for
two reasons coming from our cognitive model:
• Students may not see demonstrations as important.

• Students may not see in a demonstration what we expect them to see.
Sagredo sat in on some of my lectures in the large calculus-based class for engineers to
help me evaluate my presentation and to consider ideas that might be adaptable to his own
class. At one point in one of the lectures, I did a demonstration. The equipment had been
prepared by our superb lecture demonstration facility and was large and visible throughout
the hall. Furthermore, it worked smoothly. A few students asked questions at the end. Over-
all, I was pleased and felt it went well.
Sagredo came up to me after class. “You will never guess what happened in your demon-
stration!” he said. “Fully half the class simply stopped paying attention when you brought
out the equipment! Only the group in the front few rows and a few scattered around were
really trying to follow. Lots of students pulled out their newspapers or started talking dis-
cretely to friends!”
Since I had been concentrating on the equipment—and on the students in the first
few rows—I hadn’t noticed this. It’s plausible, though. As discussed in chapter 3, students’
expectations about the nature of learning and their goals for the class play a big role in fil-
tering what they will pay attention to in class. At the time, it was not my habit to ask exam
questions about demonstrations, so they were reasonably certain they wouldn’t be asked
about it.
My next step, then, was to change how I did my demonstrations. I began to do them
less frequently but to spend more time on each one. I tried to engage more of the class and
assured them that there would be an exam question on one of our demonstrations. In this
environment, I learned something even more striking—but not surprising given our cogni-
tive model.
Physics education researchers learned many years ago that students often think that cir-
cular motion tends to persist after the forces producing it are removed [McCloskey 1983].
In order to attempt to deal with the common naïve conception, I did the demonstration
shown in Figure 7.1. A circular ring about 0.5 m in diameter with a piece cut out of it (about
Figure 7.1 Lecture demonstration on circular motion (after [Arons

1990]). A partial circular ring lies flat on a table, and a billiard ball is
rolled around the ring.
60° worth) was laid flat on the table. The point of the demonstration was to show that a bil-
liard ball rolled around the ring would continue on in a straight line when it reached the end
of the ring.
I did the demonstration in the following series of steps in order to engage more of the
students’ attentions in what was happening.
1. I briefly reviewed the physics—circular motion and Newton’s second law.

2. I showed the apparatus and showed what I was going to do. I rolled the ball along
the ring but stopped it before it got to the edge.
3. I asked students what they expected would happen. Some expected the correct straight
line, but most expected it would continue to curve a bit. I called for discussion, and
a number of students defended one answer or another.
4. I put the answers on the board and asked for a show of hands. It was split, with a
substantial number of students supporting each answer. (No one thought it would
continue on in the circle when there wasn’t any ring holding it in.)
5. I then showed the demonstration, letting the ball roll on beyond the edge of the ring.
Then, by a lucky chance, instead of saying: “There. You see it goes in a straight line,” I
asked them what they saw. To my absolute amazement, nearly half the students claimed that
the ball had followed the curved path they expected! The other half argued that it looked to
them like a straight line. Lots of mini-arguments broke out among the students. Somewhat
nonplussed, I looked around and found a meter stick. “Let’s see if we can decide this by
looking a bit more carefully. I’ll align the meter stick along what a straight path would be—
tangent to the point where it will leave the circle—and about an inch away. If it’s going
straight, it will stay the same distance from the ruler. If it curves, it will get farther away from
the ruler as it goes.” Now, when I did it, the path was obviously straight since it remained
parallel to the ruler. It was only at this point that I got the gasp I had expected from half
the class.
Now that I know “what they need,” I could do the demo in the future using the ruler
right away. But I feel that would be a mistake. The predictions and discussions, the taking a
stand and defending their point of view, the surprise at having mis-seen what was happen-
ing—all of these contribute to the students’ engagement in and attention to the activity. Al-
though I have no hard comparative data (it would make a nice experiment), I expect the
demo we did was much better remembered than if I had simply “done it right.” On the
midsemester exam, I gave the relevant question from the Force Concept Inventory, and more
than 80% of the students gave the correct response. This is much better than the typical re-
sults from traditional instruction.
Peer Instruction/ConcepTests • 133
PEER INSTRUCTION/CONCEPTESTS
Staff: One lecturer trained in the approach (N 30–300).
Population: Introductory algebra- or calculus-based physics students (though some
ConcepTest questions are appropriate for less sophisticated populations).
Computers: None required, but one associated with a response system permits live-
time display of quiz results.
Other Equipment: Some kind of student-response system. This can be as low tech
as cards for the students to hold up or as high tech as a computer-based system with
individual wireless remote-response devices for each student.
Time Investment: Low to moderate.
Available Materials: Text with ConcepTest questions [Mazur 1997];
http://galileo.harvard.edu
Eric Mazur describes his method for increasing students’ engagement in his lectures in his
book Peer Instruction [Mazur 1997]. His method includes three parts:
1. A web-based reading assignment at the beginning of the class (see the section on JiTT
below)
2. ConcepTests during the lecture
3. Conceptual exam questions
During the lecture he stops after a five- to seven-minute segment to present a challeng-
ing multiple-choice question about the material just covered (a ConcepTest ). This question is
concept oriented, and the distractors are based on the most common student difficulties as
shown by research. Students answer the questions at their seats by either holding up a col-
ored card showing their answer or by using a device that collects and displays the collective
response on a projection screen, such as ClassTalk™ or the Personal Response System™.
Mazur then instructs the students to discuss the problem with their neighbor for two min-
utes. At the end of this period, the students answer the question again. Usually the discussion
has produced a substantial improvement. If not, Mazur presents additional material. A sam-
ple of one of Mazur’s questions is given at the top of Figure 7.2. The ConcepTest discussion
takes another five to seven minutes, breaking the lecture up into 10- to 15-minute chunks.
The response of Mazur’s students in a Harvard algebra-based class to this question is shown
in the lower half of Figure 7.2. Note that about 50% of the students start with a correct an-
swer before discussion and about 70% have the right answer after discussion. What’s more,
the fraction of students who have the right answer and are confident about it increases from
12% to 47%. This is a rather substantial learning gain for two minutes of discussion time.
Mazur suggests that a question used in this way should be adjusted so that the initial per-
centage correct is between 35 and 70%. Less than this, and there will be too few students with
the correct answer to help the others. More than that, and either you haven’t found the right
distractors or enough students know the answer that the discussion isn’t worth the class time.
Imagine holding two bricks under water. Brick A is just beneath the
surface of the water, while B is at a greater depth. The force needed
to hold brick B in place is
(a) larger than
(b) the same as
(c) smaller than
the force required to hold brick A in place.
Before discussion After discussion
100% 100%
80% 80%
Percentage
Percentage
Just guessing Just guessing
60% 60%
Not quite sure Not quite sure
40% Pretty sure 40% Pretty sure
20% 20%
0% 0%
1 2 3 1 2 3
Answer Answer
Figure 7.2 A ConcepTest question with the results before and after discussion (from [Mazur 1997]).
For all the ConcepTest questions in a given semester, Mazur found that the fraction of
correct answers invariably increased after the two-minute discussions. A plot of this result is
shown in Figure 7.3.
Finally, sensitive to the principle that students only focus on things that you test, Mazur
includes conceptual questions on every exam. His book contains reading quizzes, ConcepTests,
and conceptual exam questions for most topics in the traditional introductory physics course.
100
80
After discussion
60
40
20 % correct answers
0
0 20 40 60 80 100
Before discussion
Figure 7.3 The fraction of correct answer before and after peer discussion (from [Mazur 1997]).
Interactive Lecture Demonstrations (ILDS) • 135
INTERACTIVE LECTURE DEMONSTRATIONS (ILDS)
Population: Introductory algebra- or calculus-based physics students (though some
ILDs are appropriate for less sophisticated populations).
Computers: One required for the lecturer.
Other Equipment: LCD or other appropriately large-screen display. Some kind of
computer-assisted data acquisition device. Specific standard lecture-demonstration
equipment is required for each ILD.
Time Investment: Low.
Available Materials: Worksheets for about 25 ILDs [Sokoloff 2001].
An approach that has proven both effective and efficient is a series of interactive lecture demon-
strations (ILDs) by Sokoloff and Thornton [Sokoloff 1997] [Sokoloff 2001]. These demon-
strations focus on fundamental conceptual issues and take up a few lecture periods (perhaps
four to six) during a semester. Most use computer-assisted data acquisition to quickly collect
and display high-quality data.
In order to get the students actively engaged, each student is given two copies of a work-
sheet to fill out during the ILD—one for predictions (which they hand in at the end) and
one for results (which they keep). Giving a few points for doing the ILD and handing in the
prediction sheet (which should not be graded) is a valuable way to both increase attendance
and get some feedback on where your students are.5
Each ILD sequence goes through a series of demonstrations illustrating simple funda-
mental principles. For example, in the kinematics demonstrations, demonstrations are car-
ried out for situations involving both constant velocity and constant acceleration. The cases
with acceleration use a fan cart to provide an approximately constant acceleration. (See Fig-
ure 7.4.) The demonstrations address a number of specific naïve conceptions, including con-
fusion about signs and confusion between a function and its derivative. A fan is used to pro-
vide an acceleration rather than gravity so as not to bring in the additional confusion caused
by going to two dimensions. The specific demonstrations are:
1. Cart moving away from motion detector at constant velocity

2. Cart moving toward the motion detector at a constant velocity
3. Cart moving away from the motion detector and speeding up at a steady rate
4. Cart moving away from the motion detector and slowing down at a steady rate (fan
opposes push)
5 Sokoloffreports that even though students are told that the prediction sheets will not be graded and that they
should leave their original predictions to be handed in, some students correct their prediction sheet to show the cor-
rect answer instead of their prediction.
5. Cart moving toward the motion detector and slowing down at a steady rate (fan op-
poses push)
6. Cart moving toward the motion detector and slowing down, then reversing direction
and speeding up
In each case, the demonstrator goes through the following steps:
• Describe the demonstration to be carried out, performing it without collecting data.

• Ask the students to make and write down individual predictions on their prediction sheets
(t one minute).
• Have the students discuss the results with their neighbors and indicate their consensus
prediction on their prediction sheets (t two to three minutes).
• Hold a class discussion, putting the various predictions on the board.
• Perform the demonstration, collecting data and having the students copy the results on
their results sheet.
• Hold a brief class discussion reflecting on why the answer obtained makes sense and the
other answers have problems.
I have carried out some of these ILDs in my algebra-based physics classes. The first time
I did them, I was tempted to leave out some of the demonstrations, finding them repetitious.
After all, once they got demo 4, isn’t demo 5 obvious? When I did this, some students came
Demonstration 3: Sketch on the axes on the right +

your predictions for the velocity-time and
Velocity
acceleration-time graphs of the cart moving away

0 t
from the motion detector and speeding up at a
steady rate.
−
+
Acceleration
0 t
Figure 7.4 The apparatus and worksheet entry for a kinematics ILD.
Interactive Lecture Demonstrations (ILDS) • 137
up after class asking for the results. When I asked them for their predictions, they had the
wrong answers, having found it difficult (not at all obvious) to make the translation from the
other cases.
One place where I have found ILDs to be extremely valuable is in the class discussion
step. This offers a tremendous opportunity to change the character of the class and your in-
teraction with the students in a fundamental way. As I described in some detail in chapter 3,
many of our students have the epistemological misconception that science in general, and
physics in particular, is about the amassing of a set of “true facts.” They think that learning
scientific reasoning and sense making are a waste of time. Given this predilection, most stu-
dents are reluctant to answer a question in class if they are not convinced they have the cor-
rect answer.
In discussing the predictions for the ILDs, I encourage the class to “be creative” and to
find not just what they think might be the correct answer (probably the one given by the A
student in the front row who answered first!), but to come up with other answers that might
be considered plausible by other people (such as a roommate who is not taking the class).
This frees the students from the burden of being personally associated with the answer they
are giving and allows them to actually express what they might really believe. (I sometimes
find it necessary to give some plausible but wrong answers myself in order to get the ball
rolling.) I then ask students to try to defend each other’s answers. This changes the charac-
ter of the discussion from one that is looking for the right answer to one that is trying to cre-
ate and evaluate a range of possible answers. The focus changes from “listing facts” to build-
ing process skills.
The results of this were quite dramatic in my class. Many more students became willing
to answer (and ask) questions, and I was able to elicit responses from many more of my stu-
dents than ever before. (In a class of 165 students, about 40 to 50 students were willing to
participate in subsequent discussions.)
The evaluations of student conceptual improvement with ILDs were done by Thornton
and Sokoloff in mechanics using the FMCE. The results they reported were spectacular, with
students in classes at Tufts and Oregon improving to 70% to 90% from a starting point of
less than 20%. Of course, this result is at the primary institution, and the demonstrations
were performed by the developers or by colleagues they themselves have trained.
Secondary users have reported some difficulties with their implementation. One col-
league of mine reported implementing ILDs and obtaining no improvement on the FCI over
traditional demonstrations [ Johnston 2001].6 In my own experience with the technique, I
find it not as easy to implement effectively as it appears on the surface. With traditional
demonstrations, students often either sit back and expect to be entertained or tune out alto-
gether. With ILDs, it is essential to get the students out of that mode and into a mode where
they are actively engaging the issues intellectually. This is not easy, especially with a class that
is accustomed to passive lecturing and instructor-oriented demonstrations. A full analysis of
ILD implementation is currently under way [Wittmann 2001].
6 In this case, the lecturer is highly dynamic and entertaining. My hypothesis is that he maintained a demonstration/
listening mode in his students rather than managing to get them engaged and thinking about the problems.
JUST-IN-TIME TEACHING ( JiTT)
Population: Introductory algebra- or calculus-based physics students.
Computers: One required for the lecturer. Students require access to the web.
Other Equipment: None.
Time Investment: Moderate to high.
Available Materials: Text with questions [Novak 1999]; website http://www.jitt.org.
The Just-in-Time Teaching or JiTT approach was developed by Gregor Novak and Andy
Gavrin at Indiana University–Purdue University Indianapolis (IUPUI) and Evelyn Patterson
at the U.S. Air Force Academy. The group has collaborated with Wolfgang Christian at David-
son College to create simulations that can be used over the web.
The JiTT approach is described in the group’s book, Just-in-Time Teaching: Blending Ac-
tive Learning with Web Technology [Novak 1999]. The method is a synergistic curriculum
model that combines modified lectures, group-discussion problem solving, and web technol-
ogy. These modifications are reasonably self-standing and can be adopted by themselves or
in combination with other new methods that are described in the next two chapters.
The JiTT approach has as its goals for student learning a number of the items addressed
in chapters 2 and 3:
• Improve conceptual understanding.

• Improve problem-solving skills.
• Develop critical thinking abilities.
• Build teamwork and communication skills.
• Learn to connect classroom learning with real-world experience.
To achieve these goals, JiTT focuses on two critical cognitive principles, one from each side
of the teaching/learning gap:
• Students learn more effectively if they are intellectually engaged.

• Instructors teach more effectively if they understand what their students think and know.
These principles are implemented by using web technology to change students’ expec-
tations as to their role in the learning process and to create a feedback loop between student
and instructor. This feedback is implemented by assigning web homework “WarmUp” as-
signments before each class. The components of the process are as follows.
1. Before each lecture, specific, carefully chosen WarmUp questions are assigned and
made available on the web. The questions concern a topic that has not yet been con-
Just-In-Time Teaching (JiTT) • 139
sidered in class and that will be addressed in the lecture and class discussions and ac-
tivities. (The detailed character of these questions is discussed below.)
2. Students are expected to do the reading and consider the questions carefully, provid-
ing their best answers. They are graded for effort, not correctness. The student re-
sponses are due a few hours before class.
3. The instructor looks at the student responses before lecture, estimates the frequency
of different responses, and selects certain responses to put on transparencies (or dis-
play electronically) to include as part of the in-class discussion and activities.
4. The class discussion and activities are built around the WarmUp questions and stu-
dent responses.
5. At the end of a topic, a tricky question known as a puzzle is put on the web for stu-
dents to answer.
The authors report that for the students, thinking about the questions beforehand, seeing
their own responses as a part of class discussion, and discovering that they can solve tricky ques-
tions using what they have learned raises the level of student engagement substantially. For the
instructor, the explicit display of student difficulties provides much more feedback than is typ-
ically available. This feedback can keep the instructor from assuming too much about what the
students know and can help direct the class discussion to where it will do the most good.
A successful implementation of JiTT relies on:
1. A mechanism for delivering questions over the web and for collecting and displaying stu-
dent answers in a convenient form. You can use a number of web environments such
as WebAssign™, CAPA, and Beyond Question, or course management systems such
as BlackBoard or WebCT.
2. A set of carefully designed warm-up questions and puzzles that get to the heart of the
physics issues. The JiTT book includes examples of 29 threefold WarmUp assignments
and 23 puzzles on the topics of mechanics, thermodynamics, E&M, and optics. Many
additional JiTT materials developed by adopters and adapters are accessible via the
JiTT website.
3. An instructor with sufficient knowledge of student difficulties and with strong skills for
leading a classroom discussion. This is something that cannot be easily provided and is
the reason I have rated this method as requiring a “moderate to high” time investment.
Sagredo, although I told you at the beginning of chapter 6 that I could not provide you
any “best method” for teaching a particular physics topic, this approach allows you to learn
about specific student difficulties and to make use of what you have learned.
Running a discussion in a large lecture in such a way that many students are involved,
that the appropriate physics is covered, and that the students get to resolve their difficulties
requires substantial skill. The JiTT book includes discussions of various specific examples that
show the kinds of techniques that can be effective.
Since the entire structure of the class relies on the student responses to the WarmUp
questions and puzzles, the choice of these questions becomes critical. The JiTT book rec-
ommends that the WarmUp questions share the following characteristics:
• They are motivated by a clear set of learning objectives.

• They introduce the students to the technical terms.
• They connect to students’ personal real-world experience.
• They confront common naïve misconceptions.
• They are extendible.
The WarmUp assignments typically include three parts: an essay question, an estimation
question, and a multiple-choice question. An example is shown in Figure 7.5. Note the in-
teresting fact that some problems are stated ambiguously. Often we try to write questions in
which all the assumptions are absolutely clear. Here, for example, it is left unstated whether
the carousel is a large mechanical object in a theme park that is driven by a motor or a small
unpowered rotating disk in a children’s playground. Furthermore, even when you have envi-
sioned the situation, the first part of the essay question has no unique answer. It depends on
how you do it. This offers a good opportunity for starting a discussion.
The second part is a true estimation question as discussed in chapter 4. Not enough in-
formation is given (How fast are the planes going when they take to the air?), and informa-
tion from personal experience must be provided (How long does it take the Earth to make
one rotation?). This is also challenging since the intermediate variable required (the speed of
the Earth’s rotation) has to be connected to the personal data by a calculation.
Essay: Suppose you are standing on the edge of a spinning carousel. You step off,
at right angles to the edge. Does this have an effect on the rotational speed of the
carousel?
Now consider it the other way. You are standing on the ground next to a spinning
carousel and you step onto the platform. Does this have an effect on the rota-
tional speed of the carousel? How is this case different from the previous case?
Estimation: The mass of the Earth is about 6 1024 kg, and its radius is about
6 106 m. Suppose you built a runway along the equator and you lined up a mil-
lion 10,000 lb airplanes and had them all take off simultaneously. Estimate the ef-
fect that would have on the rotational speed of the Earth.
Multiple Choice: An athlete spinning freely in midair cannot change his
(a) angular momentum.
(b) moment of inertia.
(c) rotational kinetic energy.
(d) All of the above conclusions are valid.
Figure 7.5 A JiTT WarmUp assignment. These are distributed on the web and answered by students
before they are discussed in class.
Just-In-Time Teaching (JiTT) • 141
Hold a basketball in one hand, chest high. Hold a baseball in the other hand about
two inches above the basketball. Drop them simultaneously onto a hard floor. The
basketball will rebound and collide with the baseball above it. How fast will the
baseball rebound? Assume that the basketball is three to four times heavier than
the baseball.
The result will surprise you. Don’t do this in the house!
Figure 7.6 A JiTT puzzle.
The multiple-choice question is not at all straightforward, though I would have offered
it as a multiple-choice multiple-response question (see chapter 4), allowing the students to
pick as many of the answers as they desired. A natural error here is to choose both (a) and
(c), since both conservation of angular momentum and energy have been discussed. This
choice is not available in the form presented. The discussion of this WarmUp cluster can be
tied in class to the classic demonstration of the student with dumb-bells on the rotating chair.
A typical puzzle is given in Figure 7.6. Novak and colleagues report that most stu-
dents attempting this problem get bogged down in the algebra. They then spend a full hour
discussing this problem, using it as an opportunity to thoroughly review everything that had
been covered to that point and to discuss and build problem-solving skills.
This example nicely illustrates the difference between JiTT questions and traditional
homework problems. The goal of a JiTT question is not to evaluate students’ problem-
solving skills. In that case, you would hope that you had presented a question that most stu-
dents can answer. In constructing JiTT questions, you want a question that is not so diffi-
cult that most students are unwilling to spend any time thinking about it, but that is hard
enough that many students will not be able to complete it successfully. The primary goal of
the questions is an engaged and effective lecture discussion.
The JiTT group also includes in their approach web homework of a more standard type
and problems based on simulations. The book contains a brief introduction to creating sim-
ulations in the Physlet environment7 and a set of problems and questions that can be assigned
in conjunction with existing simulations.
The JiTT approach can be used in a variety of lecture-based classes and can readily be
combined with other techniques in recitation and laboratory.
7Physlets
is a set of programming tools using Java and JavaScript that allows the creation of simple simulations that
can be delivered on the web [Christian 2001].
CHAPTER 8
Recitation and
Laboratory-Based Methods
The most serious criticism which can be urged

against modern laboratory work in Physics is that
it often degenerates into a servile following of directions,
and thus loses all save a purely manipulative value.
Important as is dexterity in the handling and adjustment of apparatus,
it can not be too strongly emphasized
that it is grasp of principles, not skill in manipulation
which should be the primary object of General Physics courses.
Robert A. Millikan [Millikan 1903]
The recitation and the laboratory are two elements of the traditional structure that seem
ready made for active engagement. The architectural environment can be arranged to be
conducive to group work, focus on the task, and interaction. (See Figure 6.3.) Unfortu-
nately, not much is usually done with the cognitive environment to take advantage of this
opportunity. Recitations are set up with the room’s movable chairs lined up as if in a
large lecture hall (see Figure 10.3)—and the recitation leader does 95% of the talking.
Students in laboratories may sit at tables in two groups of two, but if the lab is set up
in “cookbook” style so that students can get through it quickly and without much thought,
there may be little conversation and almost no effort at sense-making.
In this chapter I discuss five environments, three for recitation and two for lab.
• The traditional recitation

• Tutorials —Materials developed by the University of Washington Physics Education
Group (UWPEG) to replace recitations by guided group concept building
• ABP Tutorials —Materials in the frame of those developed by the UWPEG but making
use of computer-assisted data acquisition and analysis (CADAA) and video technology
142
The Traditional Recitation • 143
• Cooperative Problem Solving (CPS)—An environment developed at the University of Min-

nesota to provide guidance for students to learn complex problem-solving skills in a
group-learning environment1
• The traditional lab
• RealTime Physics —A concept-building laboratory making extensive use of CADAA
Both sets of Tutorial materials and RealTime Physics are part of the Physics Suite. The
CPS materials match well with and are easily integrated with other Suite elements.
THE TRADITIONAL RECITATION
Environment: Recitation.
Staff: One instructor or assistant per class for a class of 20 to 30 students.
Population: Introductory physics students.
Computers: None.
Time Investment: Low.
The traditional recitation has an instructor (at the large research universities this is often a
graduate student) leading a one-hour class for 20 to 30 students. These sections are often tied
to the homework: students ask questions about the assigned problems, and the teaching as-
sistant (TA) models the solution on the blackboard. If the students don’t ask questions about
any particular problems, the TA might choose problems of his or her own and model those.
A brief quiz (10 to 15 minutes) may be given to make sure students attend. At Maryland,
this regime has been standard practice for decades. Sometimes, due to time pressures and a
limited number of TAs, homework grading is dropped, and the quiz is one of the homework
problems chosen at random—to guarantee that the students have to do them all, even though
they aren’t collected. The recitation becomes a noninteractive lecture in which the students
are almost entirely passive.
When I first taught the calculus-based physics class about a dozen years ago, I asked
Sagredo for his advice. “Problem solving is really important,” he responded, “so be sure your
TAs give a quiz to bring the students into the recitation on a regular basis.” I was intrigued
by the assumption implicit in this statement: that the activity was important for student learn-
ing but that without a compulsion, students would not recognize this fact.
I decided to test this for myself. I told my students that in recitation, the TAs would be
going over problems of the type that would appear on the exams. They would not be re-
quired to come, but it would help them do better on the exams. The result was the disaster
that Sagredo had predicted. Attendance at the recitations dropped precipitously. When, about
1 The CPS project has also developed a laboratory curriculum that articulates with the problem-solving recitations,
putting each laboratory exercise into the context of a problem. These laboratories are not discussed in detail here.
For more information, see [Heller 1996] and the group’s website at http://www.physics.umn.edu/groups/physed.
144 • Chapter 8: Recitation and Laboratory-Based Methods
halfway into the semester, I asked one of my TAs how his attendance was, he remarked: “It
was great last week. I actually had eight students show up [out of a class of 30].” Now and
then I stood outside one of these recitation rooms to listen to what was going on. It seemed
that there were two or three students in the group who were on top of things, had tried to
do the homework and had real questions, and were following closely. Then, there were an-
other three or four students who didn’t say a word but were writing down everything that
was said. My assumption is that they were “pattern matchers”—students who did not assume
that it was necessary to understand or make sense of the physics and felt they could get by
with memorizing a bunch of problems and then replaying them on the exam. This impres-
sion was reinforced by my interaction with these students during office hours.
The next time I taught the class I decided that since the students didn’t see recitations
as valuable to their learning, perhaps they were right. I eliminated the recitation in favor of
a group-learning concept-building activity, Tutorials [Tutorials 1998].2 I told them that in
Tutorial we would be working through basic concepts. They would not be required to come,
but it would help them do better on the exams. Interestingly enough, despite the similarity
of the instructions, the attendance results were dramatically different. The TAs reported al-
most full classes (80% to 95%) at every session.3 I don’t fully understand the psychology be-
hind this, but my first guess is that the social character of the Tutorial classes changed the
way they thought about the class. Since they were interacting with their peers, the activity
was no longer individual and they had some responsibility for being there to interact. Put an-
other way, Tutorials are like laboratories and one did not cut a lab if one could help it, in
part because it caused a serious problem for your lab partner. The traditional recitations are
more like lectures, and nobody really cared if you missed lecture.
A more interactive approach to the traditional recitation

Even if you don’t want to (or have the resources to) implement a research-based recitation
replacement such as Tutorials, in a small class of 20 to 30 you can use many techniques to
increase the students’ engagement with the material. The small-class environment provides
lots of opportunities for this engagement. Some methods include:
• Ask authentic questions —Questions that are relevant to what the students are learning
and that you expect them to answer are much more engaging than rhetorical questions
or questions that interest only one student.
• Lead a discussion—Don’t answer student questions yourself, but see if you can get a dis-
cussion going to answer the questions. Help them along now and then if needed.
• Have them work on problems together—Problems that are not assigned for homework but
that rely on an understanding of fundamental concepts can be effective and engaging.
(Problems that only rely on straightforward algebraic manipulations are not.) Having
one student from each group put their solution on the board and then having a class
2I use the word “Tutorials” with a capital T to distinguish the specific University of Washington style of lessons from
a more traditional “tutorial” in which a student is tutored—perhaps led through a lesson step by step. “Capital T”
Tutorials are a more complex activity.
3 Early morning (8 A.M.) sessions are sometimes an exception.
The Traditional Recitation • 145
discussion can be very valuable. Some of the context-based reasoning problems discussed
in chapter 4 can be effective here.
• Do fewer problems and go into them more deeply—If you do many problems quickly, it
encourages the students’ view that they need to pattern match rather than understand.
Going through a problem of medium difficulty with enough discussion that student con-
fusions are revealed may take 20 to 30 minutes.
These approaches sound easier than they are. Each one succeeds better the more you un-
derstand about where the students actually are—what knowledge they bring to the class, both
correct and incorrect, and what resources they have to build a correct knowledge structure.
The critical element is communication.
Redish’s ninth teaching commandment: Listen to your students whenever possible. Give them
the opportunity to explain what they think and pay close attention to what they say.
Helping your teaching assistants give better recitations

For instructors in charge of a group of TAs, I have some additional words of advice.
• Make sure that your TAs understand the physics —Faculty have a tendency to assume that
graduate students are well versed in introductory physics. But remember: they may be
novice TAs and have last studied introductory physics four or five years ago. Little of
what they have done since then (Lagrangians, quantum physics, Jackson problems) will
help them with the often subtle conceptual issues in an introductory class.
• Make sure that you and your TAs are on the same page—If you are trying to stress con-
ceptual issues and promote understanding, make sure that your TAs know what you are
trying to do and understand it. If you want them to use a particular method to solve a
class of problems, be sure the TAs know that you are pushing it.
• Worry about grading and administrative details—One of the easiest ways to get in trou-
ble with your students is to have different TAs grading them in different ways. A TA
who grades homework casually, giving points for effort, can produce a pattern of scores
much higher than one who slashes points for trivial math errors. This can cause diffi-
culty in assigning grades fairly and can lead to significant student anger and resentment.
These guidelines are based on my experience. One should be able to create an effective
and engaging learning environment in a class of 20 to 30 students, even without adopting a
special curriculum. However, careful research has yet to be done to see what elements are crit-
ical in producing effective learning in this situation.
Studies in other environments suggest that even when the instructor is sensitive to re-
search-determined difficulties that students have with the material, research-based instruc-
tional materials may make a big difference in the effectiveness of the instruction.4 In the next
two sections I describe three elements that can transform recitations into more effective learn-
ing environments.
4 See [Cummings 1999].

TUTORIALS IN INTRODUCTORY PHYSICS
Staff: One trained facilitator per 15 students.
Population: Introductory physics students. (All are appropriate for calculus-based
physics classes; some are also appropriate for algebra-based classes.)
Computers: Very limited use.
Other Equipment: Butcher paper or whiteboards and markers for each group of
three to four students. Occasional small bits of laboratory equipment for each group
(e.g., batteries, wires, and bulbs).
Time Investment: Moderate to substantial (one to two hour weekly training of
staff required).
Available Materials: A manual of tutorial worksheets and homework.
Perhaps the most carefully researched curriculum innovation for introductory calculus-based
physics is Tutorials in Introductory Physics, developed by Lillian C. McDermott, Peter Shaf-
fer, and the University of Washington Physics Education Group (UWPEG). Numerous Ph.D.
dissertations by students in this group have extensively investigated student difficulties with
particular topics in calculus-based physics and have designed group-learning lessons using the
research-and-redevelopment wheel. (See Figure 6.1.) References to this extensive body of work
can be found in the Resource Letter included in the Appendix of this volume. The published
materials cover a wide range of topics from kinematics to physical optics [Tutorials 1998].
Additional materials are continually being developed and refined.
In Tutorials, the traditional recitation is replaced by a group-learning activity with care-
fully designed research-based worksheets. These worksheets emphasize concept building and
qualitative reasoning. They make use of cognitive conflict and bridging, and use trained
facilitators to assist in helping students resolve their own confusions. The method can be
implemented to help improve student understanding of fundamental physics concepts in
a cost-effective manner within the traditional lecture structure [Shaffer 1992] [McDermott
1994].
Students in Tutorials work in groups of three to four with a wandering facilitator for
every 12 to 15 students. These facilitators check the students’ progress and ask leading ques-
tions in a semi-Socratic dialog5 to help them work through difficulties in their own think-
ing. (See Figure 8.5.) The structure of the classroom (Figure 6.3) reflects the different focus
in behavior expected in a Tutorial as compared to a lecture. In this, and in other inquiry-
based classes, the student’s focus is on the work (on the table) and on the interaction with
the other students in their group.
5 See
Bob Morse’s lovely little article “The Classic Method of Mrs. Socrates” to learn more about the difference be-
tween a “Socratic” and a “semi-Socratic” dialog [Morse 1994].
Tutorials in Introductory Physics • 147
The structure of Tutorials

Tutorials have the following components:
1. A 10-minute ungraded “pre-test” is given once a week (typically in lecture). This test
asks qualitative conceptual questions about the subject to be covered in Tutorial the
following week and gets the students thinking about some (usually counterintuitive)
issues.
2. The teaching assistants and faculty involved participate in a one- to two-hour weekly
training session. In this session, the TAs do the pre-test themselves and go over the
students’ pre-tests (but don’t grade them). They discuss where the students have trou-
ble and do the Tutorial in student mode, nailing down the (often subtle) physics ideas
covered.
3. Students attend a one-hour (50-minute) session. Students work in groups of three or
four and answer questions on a worksheet that walks them through building quali-
tative reasoning on a fundamental concept.
4. Students have a brief qualitative homework assignment in which they explain their
reasoning. This helps them bring back and extend the ideas covered in the Tutorial.
It is part of their weekly homework, which in most cases also includes problems as-
signed from the text.
5. A question emphasizing material from Tutorials is asked on each examination. (See
Redish’s sixth teaching commandment!)
Tutorials often focus on important but subtle points

At the University of Washington, tutorial worksheets are developed over a period of many
years using the research-redevelopment cycle. The UWPEG has a highly favorable situation
for curriculum development at the University of Washington—a large research group of grad-
uate students and postdocs, continued support for research and development over many years,
and an educational environment in which every term of the calculus-based physics course is
taught using Tutorials four times a year. As a result, the UW Tutorials are highly refined and
very carefully thought out and tested. Although one might think one sees some obvious “fixes,”
I recommend that they not be changed lightly.
When we first introduced Tutorials at the University of Maryland, Sagredo was lectur-
ing one of the sections in which we were testing tutorials. He suggested that the vector ac-
celeration activity shown in Figure 8.1 be replaced by motion on a circle. “After all,” he com-
mented, “circular motion is much simpler than elliptical motion, so they should understand
it better.”
Sagredo’s comment misses the point of the activity. Reif and Allen have demonstrated
[Reif 1992] that students often don’t internalize the concept of vector acceleration well at all.
They try to memorize formulas that will allow them to solve problems without struggling to
make sense of the fundamental concepts. In the case of acceleration, the students needed to
learn to think of vector acceleration through a process—looking at velocity vectors at nearby
times and seeing how they changed. The activity in the UW tutorial is carefully designed to
Acceleration vectors for constant speed
Suppose that the object in part I is moving around the track at uniform speed.
• Draw vectors to represent the velocity at two points on the track that are relatively
close together. (Draw your vectors LARGE.)
• Label the two points C and D.
• On a separate part of your paper, copy the velocity vectors vC and vD.
• From these vectors, determine the change in velocity vector, ∆v.
i. How does the angle formed by the head of vC and the tail of ∆v compare to 90°?
(“Compare” in this case means “is it less than, greater than, or equal to 90°?”)
As point D is chosen to lie closer and closer to point C, what happens

to the above angle? Explain how you can tell.
What happens to the magnitude of ∆v as point D is chosen to lie closer

and closer to point C?
ii. How would you find the acceleration at point C?
Figure 8.1 A sample activity from a University of Washington Tutorial [Tutorials 1998].
be sufficiently general (not a circle, not an ellipse) so that the students can’t pattern match to
something in the book, but sufficiently specific (moving at a constant speed but with chang-
ing direction, later with changing speed) to force them to focus on the process of construct-
ing the acceleration. Following Sagredo’s well-meaning advice would have completely under-
mined the carefully designed learning activity.
Should you post solutions to Tutorial pre-tests and homework?

The UW Tutorials often rely on the cognitive conflict method discussed in chapter 2. In this
approach, situations are presented that cue common student difficulties revealed by research.
The facilitators then help those students who show the predicted difficulties work through
their ideas themselves. McDermott refers to this process as elicit/confront/resolve [McDermott
1991]. The pre-tests often raise questions that appear to be straightforward and many (if not
most) of the students miss. Note that the pre-tests should not be gone over in class, nor should
the results be posted. The point of the pre-tests is to get the students thinking about the is-
sues. They then confront these issues for themselves during the tutorial session. Giving them
the answers short-circuits the learning activity.
Sagredo was worried about this and after a few weeks of Tutorials, asked his students in
lecture whether they wouldn’t like to have the answers to the pre-tests posted. The result was
Examine the leads to the wire coil so that you understand which portion of the wire has
been stripped of the insulating enamel coating.
• For what orientations of the coil will there be a current through it due to the battery?
• Check your answer by closing the switch and observing the deflection of the
ammeter as you rotate the coil manually through one complete revolution.
• Hold one pole of the magnet close to the coil. Close the switch. If the coil does not
begin to spin, adjust the location of the magnet or gently rotate the coil to start it
spinning.
Use the ideas that we have developed in class to explain the motion of the wire coil.
(The questions on the following page may serve as a guide to help you develop an
understanding of the operation of the motor.)
Figure 8.2 An activity from a UW Tutorial involving simple equipment [Tutorials 1998].
lukewarm. One of the students spoke up and said, “Well, we just go over the answers in Tu-
torial the next week so we get it there.”
The answers to the Tutorial homework may be a different story. The Tutorial homework
is supposed to be a reasonably straightforward extrapolation of what was done in Tutorials in
order to provide reinforcement at a later time. In some of my classes, this did not appear to
be a problem, especially when most groups had a chance to finish the Tutorials. In classes
where the Tutorials often were not finished (e.g., in my algebra-based classes), students found
the Tutorial homework more difficult. When I discovered that some of my graduate assistants
had gotten the answers wrong on the Tutorial homework, I decided to provide solutions.
What does it take to implement Tutorials?

The UW Tutorials focus on fundamental concepts and low implementation costs. Although
they occasionally call for a few items of easily obtained equipment (batteries and bulbs, mag-
nets, compasses), most are done with paper and pencil as in the example shown in Figure 8.1.
Some of the activities, however, involve well-designed and exciting mini-labs. (See, for example
Figure 8.2.) The largest investment required is that someone has to become an expert in the
method and provide training for the facilitators and someone has to manage the operation.6
The tutorial and homework sheets are available for purchase by the students [Tutorials
1998]. Pre-tests, sample examination problems, and equipment lists are available with the in-
structor’s guide.
Tutorials produce substantially improved learning gains

Tutorials have been extensively researched and tested by the University of Washington group
and by others. Many of the UWPEG’s publications over the past decade have been research
associated with the development of specific Tutorials.
We carried out a test of a secondary implementation of Tutorials at the University of
Maryland using the FCI pre and post in the first semester of engineering physics [Redish 1997].
To see the range of variation that arose from different lecturers, Saul, Steinberg, and I gave the
FCI to 16 different lecture sections involving 14 different professors. Seven of the sections used
traditional recitations, and nine used Tutorials. The classes chosen to use Tutorials were se-
lected at random. Two professors taught twice, once with Tutorials and once without.
The fraction of the possible gain,7 g, attained averaged 0.20 in the lecture classes and
0.34 in the Tutorial classes. Each of the professors who taught with and without Tutorials
each had better scores with Tutorials by 0.15. (One of these professors taught with Tutorials
first, one with recitation first.) A histogram of the results is shown in Figure 8.3. Every lec-
ture section that used Tutorials achieved a higher value of g than every section that used recita-
tions. (A later class with an award-winning professor achieved a gain of 0.34 without Tuto-
rials, higher than our lowest gain with Tutorials but lower than most of the Tutorial-based
classes.)
Changing recitations to Tutorials doesn’t hurt problem solving

In their dissertations at Maryland, Jeff Saul and Mel Sabella studied problem solving in our
tutorial/recitation comparison. In most cases, there was little or no difference observed be-
tween the two groups on traditional exam problems. On a few problems, the tutorial stu-
dents did dramatically better than those in recitations. The interesting cases are those where
the tutorial did not specifically cover the kind of example in the problem but appeared to
help students build a functional mental model.
An example of such a problem is shown in Figure 8.4, and the results at Maryland are
shown in Table 8.1 [Ambrose 1998]. The problem is trickier than it looks. Students who
memorize equations tend to memorize them in the order given in the book, and the one for
the position of the bright fringes is always given first. The problem, however, asks for the po-
sition of the dark fringe. A large fraction of the students in the recitation class simply pulled
out the bright-fringe formula and got an answer off by a factor of 2.
6 At Maryland, we found that the extra costs needed to run Tutorials for 600 students corresponded to about one-
half of a teaching assistant—the person needed to organize and manage the operation.
7 See the discussion of g in chapter 5 for a definition.
5
Recitation
4 Tutorial
# of classes
3
0
.00-.02
.04-.06
.08-.10
.12-.14
.16-.18
.20-.22
.24-.26
.28-.30
.32-.34
.36-.38
.40-.42
.44-.46
g (fraction of possible gain)
Figure 8.3 Fraction of the possible gain attained by engineering physics students at the University of
Maryland in classes taught with traditional recitations (dark) and tutorials (light).
A satisfyingly large fraction of the students in the Tutorial class actually reasoned their
way to the answer using a path-length argument, showing that they could call on an under-
lying mental model to construct a correct result. I was particularly impressed with this result
since the Tutorials we used do not explicitly consider this problem but focus on building the
concept of path length and its role in interference.
Students need to get used to Tutorials

In introducing Tutorials to a class, you should be aware of possible attitudinal difficulties. As
discussed in chapter 3, students bring to their physics classes expectations about the type of
knowledge they will be learning in class and what they have to do to get it. Engineering stu-
dents (especially those who have taken AP physics in high school) may have a strong expec-
tation that what they are supposed to learn in a physics class are equations and how to pro-
duce numbers. The idea of “concepts” and even the idea of “making sense” of anything in
physics may be foreign to them. These students can at first be quite hostile to the idea be-
hind Tutorials. Some of the better students think Tutorials are trivial (despite making nu-
merous errors in their predictions). Others may be accustomed to operating in a competitive
Light with 500 nm is incident on two narrow slits separated by d 30 m.

An interference pattern is observed on a screen a distance L away from the slits.
The first dark fringe is found to be 1.5 cm from the central maximum. Find L.
Figure 8.4 A problem on which Tutorial students performed significantly better than recitation stu-
dents [Ambrose 1998].
TABLE 8.1 Results on Problem Given to Recitation and Tutorial Classes

Recitation Tutorial
Example (N 165) (N 117)
L 1.8 m D d sin /2 16% 60%
(correct) sin y /L
L 0.9 m y mL/d 40% 9%
Other L 5.0 107 m 44% 31%
rather than a cooperative framework and may not like “having to explain their answers to
dummies.” (I’ve gotten this comment even after a session when one of the “dummies” asked
a probing question that helped that overconfident self-categorized “top student” correct a se-
rious error in his thinking.) Once both the faculty and the student body come to accept Tu-
torials as a normal part of the class, Tutorials tend to be rated as one of the most valuable
parts of the class.
Given that conceptual learning and qualitative reasoning may be new to many of the
students in an introductory physics class, the introduction of Tutorials needs to be done care-
fully. I have had the most success when I have integrated the Tutorial approach fully into my
lectures and tied qualitative reasoning to my problem-solving examples. Exams that contain
a “Tutorial question” are a minimum necessity. Exams in which every question blends Tuto-
rial ideas with problem solving are even more effective in helping students understand the
value of concepts and qualitative thinking.
ABP TUTORIALS
Population: Introductory calculus-based physics students. (Many of the tutorials are
also appropriate for algebra-based classes.)
Computers: One for every three to four students.
Other Equipment: Butcher paper or whiteboards and markers for each group of
three to four students. Occasional small bits of laboratory equipment for each group
(e.g., batteries, wires, and bulbs). Computer-assisted data acquisition device; various
programs and simulations including Videopoint and EM Field.
Time Investment: Moderate to substantial (one to two hours weekly training of
staff required)
Available Materials: A set of tutorial worksheets, pre-tests, and homework. [ABP-
Tutorials] These tutorials and instructions for their use are available on the web at
http://www.physics.umd.edu/perg/.
ABP Tutorials • 153
Although the UWPEG Tutorials cover a wide range of topics, they strongly focus on the is-
sue of qualitative reasoning and concept building. In addition, the UWPEG made the choice
to make Tutorials as easy to implement as possible, so they rely on very little (and very in-
expensive) equipment. In addition, the UWPEG Tutorials are designed so that they can be
reasonably successful in helping students build fundamental concepts even if concept build-
ing is not significantly supported elsewhere in the course (lecture, laboratory, homework prob-
lems). One difficulty with such a situation is that students tend to develop independent
schemas for qualitative and quantitative problem solving and to only occasionally (as dis-
cussed in the section on Tutorials) cross qualitative ideas over to help them in solving quan-
titative problems [Kanim 1999] [Sabella 1999].
ABP Tutorials are mathematically and technologically oriented

The University of Maryland Physics Education Research Group (UMdPERG), as part of the
Activity-Based Physics (ABP) project, developed a supplementary set of Tutorials that are
based on a different set of assumptions [Steinberg 1997]:
1. We assume that conceptual learning is being integrated throughout the course—in
lecture, homework, and laboratories—so that Tutorials are not the sole source of con-
ceptual development.
2. We assume that quantitative problem solving is a significant goal of the class.
3. We assume that reasonable computer tools are available for use in Tutorials.
Given these assumptions, the structure of Tutorial lessons can be changed somewhat. They
can focus more on relating conceptual and mathematical representations and on building
qualitative to quantitative links. In this set of lessons, computers are used for taking data,
displaying videos, and displaying simulations. For example, in Figure 8.5, a facilitator
Figure 8.5 Interactive computer-based tutorial on Newton’s law. Students are interacting with the fa-
cilitator (standing), who asks Socratic-dialog questions to focus their inquiries.
(standing) is shown talking to a group of students who are using a fan cart on a Pasco track
with a sonic ranger detector to study Newton’s second law.
While some of the lessons are new, others have been adapted to the Tutorial framework
from RealTime Physics laboratories and from Workshop Physics activities. These include
• Discovering Newton’s third law using two force probes on Pasco carts
• Exploring the concept of velocity by walking in front of a sonic ranger
• Exploring oscillatory behavior by combining a mass on a spring hanging from a force
probe with a sonic ranger
New lessons include such topics as
• Tying the concepts of electric field and electrostatic potential to their mathematical rep-
resentations using the software EM Field [Trowbridge 1995]
• Building an understanding of the functional representation of wave propagation using
video clips of pulses on springs and the video analysis program Videopoint™ [Luet-
zelschwab 1997]
• Building an understanding of the oscillatory character of sound waves and the meaning
of wavelength using video clips of a candle flame oscillating in front of a loudspeaker
and the video analysis program Videopoint™ (See Figure 8.6.)
Concept learning can be tied to the use of math

The example of sound waves gives an interesting example of how concept learning can be
tied to mathematical concepts by using media effectively.
Figure 8.6 A frame from a

video used in an ABP Tutorial.
A low-frequency sound wave
(10 Hz) emitted by the speaker
causes the candle flame to os-
cillate back and forth. Students
use Videopoint™ to measure
the frequency of the oscillation.
ABP Tutorials • 155
Students studying the topic of sound in the traditional way often construct a picture of
a sound wave that treats the peaks of the wave (or pulse) as if it were a condensed object pushed
into the medium rather than as a displacement of the medium [Wittmann 2000]. Wittmann
refers to these two mental models as the particle pulse model (PP) and the community consen-
sus model (CC). One clue that a student is using a PP model is that the student assumes that
each pulse of sound that passes a floating dust particle “hits” it and pushes it along.
Alex: [The dust particle] would move away from the speaker, pushed by the wave, pushed by
the sound wave. I mean, sound waves spread through the air, which means the air is actu-
ally moving, so the dust particle should be moving with that air which is spreading away
from the speaker.
Interviewer: Okay, so the air moves away—
A: It should carry the dust particle with it.
I: How does [the air] move to carry the dust particle with it?
A: Should push it, I mean, how else is it going to move it? [sketches a typical sine curve] If
you look at it, if the particle is here, and this first compression part of the wave hits it, it
should move it through, and carry [the dust particle] with it.
...
I: So each compression wave has the effect of kicking the particle forward?
A: Yeah.8
In his thesis research, Wittmann found that most students in engineering physics used
the PP model most of the time but used the CC model in some circumstances. One ABP
Tutorial developed to deal with this issue uses a video of a flame in front of a loudspeaker.
A dust particle is located in front of a silent loudspeaker

(see figure). The loudspeaker is turned on and plays
a note at a constant (low) pitch. Which choice or
combination of the choices a–f (listed below) can describe
the motion of the dust particle after the loudspeaker is
turned on? Circle the correct letter or letters. Explain.
Dust particle
Loudspeaker
Possible responses for question 2:
a) The dust particle will move up and down.
b) The dust particle will be pushed away from the speaker.
c) The dust particle will move toward and away from the speaker.
d) The dust particle will not move at all.
e) The dust particle will move in a circular path.
f) None of these answers is correct.
Figure 8.7 MCMR problem used to probe mental models students use in envisioning sound.
8 Dialog quoted from [Wittmann 2001a].

TABLE 8.2 Student Performance on Sound Wave Questions Before, After Traditional
Lecture, and After Additional Modified Tutorial Instruction
Time during
Semester Before All Post-Lecture,
MM used Instruction (%) Post-Lecture (%) Post-Tutorial (%)
CC (longitudinal oscillation) 9 26 45
Other oscillation 23 22 18
PP (pushed away linearly or sinusoidally) 50 39 11
Other 7 12 6
Blank 11 2 21
(See Figure 8.6.) Using cognitive conflict methods, the lesson has students predict the flame’s
behavior when the speaker is turned on. Students then track how the flame moves back and
forth and create a graph of the tip’s oscillatory motion. Then they consider what will happen
when a sound wave passes a chain of separated flames in order to build an understanding of
the idea of relative phase and wavelength.
An exam question used to test the students’ responses to this lesson is shown in Figure
8.7, and the results for traditional and Tutorial instruction are presented in Table 8.2. Data
are matched (N 137 students). The large number of blank responses in the post-all in-
struction category is due to the number of students who did not complete the pre-test on
which the question was asked. The results show that Tutorials are a significant improvement
over traditional instruction.9
COOPERATIVE PROBLEM SOLVING (CPS)
Staff: One instructor or assistant per class trained in the approach for any number
of students (N 20–30); a second facilitator is helpful in larger classes.
Population: Introductory calculus-based physics students. (Many of the problems are
also appropriate for algebra-based classes.)
Computers: None.
Time Investment: Moderate to substantial.
Available Materials: Manuals of problems for students and an instructor’s guide.
Available from the group website at http://www.physics.umn.edu/groups/physed.
9 The development environment at Maryland does not match the ideal one at Washington, so these Tutorials have
only been through two to four cycles of development compared to the eight to ten typically carried out at Wash-
ington. As a result, they are not as refined—and not as effective.
Cooperative Problem Solving • 157
Over the past decade or so, Pat and Ken Heller at the University of Minnesota and their col-
laborators have developed a group-learning problem-solving environment in which students
work together in recitation on problems they have not previously seen [Heller 1992]. Their
work is based on the generalized studies of the Johnsons and their collaborators on the ef-
fectiveness of group learning [ Johnson 1993].
Cooperative Problem Solving relies on context-rich problems

The problems the Minnesota group have developed are context rich; that is, they involve re-
alistic situations, may contain incomplete data, and may require the students to pose a part
of the problem themselves. (See Figure 8.8.) The problems are intended to be too difficult
for any individual student to solve but not too hard for a group of students of mixed ability
to solve in about 15 to 20 minutes when working together. Groups are formed to include
students of varying ability, and students may be assigned specific (and rotating) roles to play
in each group.
The Minnesota group’s context-rich problems have a number of general characteristics that
encourage appropriate thinking. These characteristics may be difficult for a novice problem-
solver to handle individually, and they facilitate discussion. They include the following:
1. It is difficult to use a formula to plug in numbers to get an answer.

2. It is difficult to find a matching solution pattern to get an answer.
3. It is difficult to solve the problem without first analyzing the problem situation.
4. It is difficult to understand what is going on in this problem without drawing a pic-
ture and designating the important quantities on that picture.
5. Physics words such as “inclined plane,” “starting from rest,” or “projectile motion”
are avoided as much as possible.
6. Logical analysis using fundamental concepts is reinforced.
A friend of yours, a guitarist, knows you are taking physics this semester and asks
for assistance in solving a problem. Your friend explains that he keeps breaking
repeatedly the low E string (640 Hz) on his Gibson “Les Paul” when he tunes up
before a gig. The cost of buying new string is getting out of hand, so your friend
is desperate to resolve his dilemma. Your friend tells you that the E string he is
now using is made of copper and has a diameter of 0.063 inches. You do some
quick calculations and, given the neck of your friend’s guitar, estimate the the wave
speed on the E string is 1900 ft/s. While reading about stringed instruments in
the library, you discover that most musical instrument strings will break if they
are subjected to a strain greater than about 2%. How do you suggest your friend
solve his problem?
Figure 8.8 Sample of a context-rich problem from the Minnesota CPS method [Heller 1999].
The problems are written so as to require careful thinking.
• The problem is a short story in which the major character is the student. That is, each
problem statement uses the personal pronoun “you.”
• The problem statement includes a plausible motivation or reason for “you” to calculate
something.
• The objects in the problems are real (or can be imagined)—the idealization process oc-
curs explicitly.
• No pictures or diagrams are given with the problems. Students must visualize the situa-
tion by using their own experiences.
• The problem solution requires more than one step of logical and mathematical reason-
ing. There is no single equation that solves the problem.
These types of problems change the frame: students cannot simply “plug-and-chug” or
pattern match. They have to think about the physics and make decisions as to what is rele-
vant. This strongly encourages the group to try to make sense of the problem rather than
simply come up with the answer.
The Minnesota group does not simply drop these harder problems on their students.
They develop an explicit problem-solving strategy, and they help the students apply it when
they get stuck. The broad outlines are illustrated in Figure 8.9.10
Although it’s hard to create such problems, when you get one, the effect of the group
interaction can be quite dramatic. But the whole idea of solving problems in a group may be
difficult—for the TAs as much as for the students. One year I prepared some of these prob-
lems and handed one out to my TAs on transparencies each week, in order to encourage them
to begin some group work instead of lecturing to the students. One TA, having had no ex-
perience with group work herself, decided not to follow my instructions. Instead of assign-
ing the problem as group work, she presented it as a quiz at the beginning of the class. When
after 10 minutes most of the students said they had no idea how to begin, she let them use
their class notes. When after an additional five minutes they still were not making progress,
she let them open their texts. After 20 minutes, she collected and graded the quiz. The re-
sults were awful, the average being about 20%. The students in her class complained that
“the quizzes were too hard.” When I questioned her about what she had done, she replied
that “If they work together, you don’t know who’s responsible for the work.” She had mis-
taken an activity which I had intended to serve a teaching purpose for one meant to serve as
an assessment. In other sections, many groups solved the problem successfully.
Group interactions play a critical role

Sagredo had some sympathy for my TA. “If they work together, you’ll just get to see the work
of the best student. The weaker students will just go along for the ride,” he complained. The
10 This strategy is an elaboration of the strategy found in Polya’s famous little book, How to Solve It [Polya 1945].
The Minnesota group found that using Polya’s strategy directly was too difficult for the algebra-based class and
that some intermediate elaborations were required [Heller 1992]. Polya’s strategy is: (1) understanding the problem,
(2) devising a plan, (3) carrying out the plan, and (4) looking back.
Cooperative Problem Solving • 159
• construct a mental image

FOCUS on the PROBLEM • sketch a picture
“What’s going on?” • determine the question
• select a qualitative approach
• diagram space-time relations

• define relevant symbols
DESCRIBE THE PHYSICS • declare a target quantity
• state quantitative relationships
from general principles and
specific constraints
• choose a relationship
containing the target quantity
• cycle: (more unknowns? choose
PLAN THE SOLUTION new relation involving it)
• solve and substitute
• solve for target
• check units
• put in numbers with units

• fix units
EXECUTE THE PLAN • combine to calculate a number
• simplify expression and units
• check answer properly stated

• check for reasonableness
EVALUATE THE ANSWER • review solution
• check for completeness
Figure 8.9 Structure of the problem-solving method used by the Minnesota group (simplified and
condensed somewhat) [Heller 1999].
Minnesota group has shown that this is not the case, and they have developed techniques to
improve group interactions.
The work of the group is better than the work of the best student in it
In order to evaluate the success of groups compared to the success of the best individual in
the group, the Minnesota group compared individual and group problem-solving success
[Heller 1992]. Since you can’t give the same problems to the same students in different con-
texts and compare the results, they developed a scheme for determining problems that had
approximately the same level of difficulty. They classified problem difficulty by considering
six characteristics.
1. Context: Problems with contexts familiar to most students (through direct experience,
newspapers, or TV) are less difficult than those involving unfamiliar technical con-
texts (such as cyclotrons or X-ray signals from pulsars).
2. Cues: Problems containing direct cues to particular physics (mention of force or “ac-
tion and reaction”) are less difficult than those for which the physics must be inferred.
3. Match of given information: Problems with extraneous information or information
that needs to be recalled or estimated are more difficult than those where the infor-
mation provided precisely matches the information needed.
4. Explicitness: Problems where the unknown required is specified are easier than those
for which it has to be invented.
5. Number of approaches required: Problems that only need one set of related principles
(e.g., kinematics or energy conservation) are less difficult than those requiring more
than one set of such principles.
6. Memory load: Problems that require the solution of five equations or fewer are easier
than those requiring more.
For each problem they assigned a difficulty value of 0 or 1 on each of these characteristics
and found that the problem score was a good predictor of average student performance.
In order to test whether the groups were operating effectively to produce better solutions
or whether they simply represented the work of the group’s best student, they tested their
students using both group and individual problems. Problems were matched as to level of
difficulty on the scale described above. Over six exams in two terms, the groups averaged 81
(N 179), while the best-in-group individual had an average of 57. These results were con-
sistent over the different exams and classes and strongly suggest that the groups are performing
better than the best individual in the group.11 By now, the group has developed a much more
detailed structure for identifying the difficulty level of a question. See the group website for
details.
Techniques for improving group interactions

The Minnesota group has studied the dynamics of group interactions in CPS and has devel-
oped a number of recommendations:
• Assign roles: In order to combat the tendency of students to select narrow roles during
group activities and therefore limit their learning, the Minnesota group assigns roles to
students in the group: manager, explainer, skeptic, and recordkeeper. These roles rotate
throughout the semester.
• Choose groups of three: Groups of two were not as effective in providing either concep-
tual or procedural knowledge as groups of three or four. In groups of four, one student
sometimes tended to drop out—either a timid student unsure of him/herself, or a good
student tired of explaining things.
11 Studentswere given unlimited time to complete the final exam. Their comparative study of incomplete problems
in the midsemester (time limited) and final exams shows that time considerations do not upset this conclusion [Heller
1992].
The Traditional Laboratory • 161
• Assign groups to mix ability levels: Groups with strong, medium, and weak students per-
formed as well as groups containing strong students only. Often, the questions asked by
weaker students helped strong students identify errors in their thinking. Groups of uni-
formly strong students also tended to overcomplicate problems.
• Watch out for gender problems: Groups of two males and one female tend to be domi-
nated by the males even when the female is the best student in the group.
• Help groups that are too quick to come to a conclusion: This can occur when a dominant
personality railroads the group or because of a desire to quickly accept the first answer
given. Some groups try to quickly resolve disagreements by voting instead of facing up
to their differences and resolving them.
I have noted this last problem in Tutorials as well. In both cases, facilitators can help get
them back on track, encouraging them to reconsider and resolve discrepancies. The Min-
nesota group suggests that group testing can help with this problem.
In Jeff Saul’s dissertation, he studied four different curricular innovations including co-
operative problem solving [Saul 1998]. He observed both the University of Minnesota’s im-
plementation in calculus-based physics and a secondary implementation at the Ohio State
University. Pre-post testing with the FCI indicated that CPS was comparable to Tutorials in
producing improvements in the student’s conceptual understanding of Newtonian mechan-
ics. (See Figure 9.4.) This is interesting since CPS focuses on quantitative rather than quali-
tative problem solving.
Unfortunately, Saul found no significant gains on the MPEX survey.12 So even though stu-
dents appeared to improve their conceptual knowledge (and their ability to use that knowledge),
their conscious awareness of the role of concepts did not seem to improve correspondingly.
THE TRADITIONAL LABORATORY
Environment: Laboratory.
Staff: One instructor or assistant per class trained in the approach for a class of 20
to 30 students.
Population: Introductory physics students.
Computers: If desired, one for every pair of students.
Other Equipment: Laboratory equipment.
Time Investment: Medium.
The laboratory is the single item in a traditional physics course where the student is expected
to be actively engaged during the class period. Unfortunately, in many cases the laboratory
has turned into a place to either “demonstrate the truth of something taught in lecture” or
12 See the discussion of the MPEX in chapter 5.

to “produce a good result.” The focus in both of these cases is on the content and not on
what might be valuable for a student to learn from the activity. In the United States, “cook-
book” laboratories—those in which highly explicit instructions are given and the student
doesn’t have to think—are common. They are unpopular with students and tend to produce
little learning. Some interesting “guided-discovery labs” have been developed in the past few
years that appear to be more effective.
Despite some interesting research on learning in laboratories in the early years of PER
(e.g., [Reif 1979]) and a few recent studies (e.g., [Allie 1998] and [Sere 1993]), there has
been little published research on what happens in university physics laboratories.
Goals of the laboratory

One can imagine a variety of goals for a laboratory:
• Confirmation—To demonstrate the correctness of theoretical results presented in lecture.

• Mechanical skills—To help students attain dexterity in handling apparatus.
• Device experience—To familiarize students with measuring tools.
• Understanding Error—To help students learn the tools of experiment as a method to
convince others of your results: statistics, error analysis, and the ideas of accuracy and
precision.
• Concept building—To help students understand fundamental physics concepts.
• Empiricism—To help students understand the empirical basis of science.
• Exposure to research—To help students get a feel for what scientific exploration and re-
search are like.
• Attitudes and expectations —To help students build their understanding of the role of in-
dependent thought and coherence in scientific thinking.
This is a powerful and daunting list. Most laboratories have at first rather limited and
practical goals—to satisfy the requirements of the engineering school or to qualify for pre-
medical certification. In implementation, most laboratories only explicitly try to achieve the
first two or three goals. Sometimes understanding error is an explicit goal, but in my expe-
rience, traditional laboratories fail badly at this goal. Students go through the motions but
rarely understand the point. Extensive research on this issue is badly needed.
Often less happens in traditional labs than we might hope

In my research group’s observation of traditional laboratories, one result is clear: the dialogs
that take place are extremely narrow [Lippmann 2002]. Our videotapes show students spend-
ing most of the period trying to read the manual and figure out what it wants them to do.
The students make little or no attempt to synthesize in order to get an overview of what the
point of the lab is. Almost all of the discussion concentrates on the concrete questions of how
to configure, run, and get information from the apparatus. There is little or no discussion of
The Traditional Laboratory • 163
the purpose of the measurement, how it will be used, the physics to be extracted, or the lim-
itations of the measurements. The students are so focused on achieving the “paper” goals of
the lab—getting numbers to be able to construct lab reports—that the learning goals appear
totally lost.
Of course, one might hope that they “get the numbers” in the lab and then “think about
them” outside of class. This may be the case, but I suspect it is a pious hope. Students rarely
have the skills to think deeply about experiments. This is where they need the help and guid-
ance of an instructor, and they don’t get it if this activity is carried out outside of class (or
with instructors who can’t handle the serious pedagogy needed).
A more interactive approach to the traditional laboratory

Some of my colleagues and TAs have experimented with variations in the traditional labora-
tory in order to get the students more intellectually engaged. From this anecdotal evidence,
I extract a few tentative guidelines. I sincerely hope that in a few years, educational research
will be able to “put legs” under these speculations.
• Make it a “class” through discussion—Often students in lab speak only to their lab partner.
There is little sharing of results or problems with other students in the class. An overall
class discussion at the beginning and end of the class might increase the engagement.
• Take away the lab manual —Having a step-by-step procedure may guarantee that most
students complete the lab but undermines important learning goals. Pat Cooney at
Millersville University has had success with simply writing the task on the board and
having students figure out what they have to do.
• Start the class with a planning discussion —Most students do not spontaneously relate the
broad goals of the lab to the details of the measurements. Having them think about these
issues before beginning their measurements is probably a good idea. Bob Chambers at
the University of Arizona has had good results with two-week labs in which the students
use the first week to plan the experiment and the second week to carry it out.
• Occasionally ask them what they’re doing and why —Students frequently get lost in the de-
tails of an activity and can get off on the wrong track. Asking them perspective ques-
tions (“What are you doing here? What will that tell you? What could go wrong?”) might
help them make the connection to the purpose of the experiment.
• Share results—Arranging labs so that there is some time for discussion and sharing of re-
sults at the end of the lab might help identify problems and give students a better idea
of the meaning of experimental uncertainty.
The laboratory is the traditional instructional environment that is, in principle, best set
up for independent active-engagement learning in line with our cognitive model of learning.
Much more research will need to be done in order to figure out what learning goals can be
effectively accomplished in the laboratory environment and how.
REALTIME PHYSICS
Environment: Laboratory.
Population: Introductory calculus-based physics students.
Computers: One for every two to four students.
Other Equipment: An analog-to-digital converter (ADC). Probes needed for the
ADC include motion detector (sonic ranger), force probes, pressure and tempera-
ture probes, current and voltage probes, and rotary motion probe. Low-friction carts
and tracks required for the mechanics experiments.
Time Investment: Low to moderate.
Available Materials: Three published manuals of laboratory worksheets for Me-
chanics (12 labs), Heat and Thermodynamics (6 labs), and electric circuits (8 labs)
[Sokoloff 1998–2000]. Laboratories in electricity and optics are under development.
An instructor’s guide is available on-line to registered instructors at http://www.
wiley.com/college/sokoloff-physics/.
Sokoloff, Thornton, and Laws have recently combined to develop a new series of mechanics
laboratories that can be used in a traditional lecture/lab/recitation teaching environment. They
make heavy use of computer-assisted data acquisition and the results of research on student
difficulties.
RTP uses cognitive conflict and technology to build concepts

The primary goal for these laboratory exercises is to help students acquire a good under-
standing of a set of related physics concepts [Thornton 1996]. Additional goals include pro-
viding students with experience using microcomputers for data collection, display, and anal-
ysis, and enhancing laboratory skills. The primary goal has been extensively tested by the
designers and by other researchers using standardized evaluation surveys. Significant gains ap-
pear to be possible. (These results are discussed in detail at the end of this section.)
The critical tool in these laboratories is an analog-to-digital converter (ADC) connected
to a computer. Many different probes can connect to these ADCs and provide the student
with graphs of a wide variety of measured and inferred variables. Our senses do not provide
us with direct measures of many of the quantities that are critical to an understanding of fun-
damental physics concepts. Our brains easily infer position and change in position, but in-
ferring speed from visual data seems to be a learned skill (and one most people who have ex-
perience crossing streets learn effectively). Acceleration, on the other hand, seems to be quite
a bit more difficult. Our brains easily infer rate of heat flow to the skin but are hard pressed
to distinguish that from temperature. The computer probes allow live-time plots13 of posi-
tion, temperature, pressure, force, current, voltage, and even of such complex calculated
13 “Live-time” in this context means that there is no noticeable time delay between the event being measured and
the display of the data.
RealTime Physics • 165
Figure 8.10 Analog-to-digital converters from Vernier and Pasco. Either box connects to the com-
puter’s serial port. A variety of probes can be connected to the box’s front (shown).
variables as velocity, acceleration, and kinetic energy. The ADCs from Pasco and Vernier are
shown in Figure 8.10.
The authors combine pedagogically validated methods such as cognitive conflict, bridg-
ing, and the learning cycle (exploration/concept introduction/concept application) with the
power of computer-assisted data acquisition to help students re-map their interpretation of
their experience with the physical world.14
RTP relies on psychological calibration of technology

When I first told Sagredo about the microcomputer use in the laboratory, he complained.
“But if they don’t understand how a measurement is made, they don’t really understand what
it means.” This may be true, Sagredo. I am fairly certain that my introductory students who
use the sonic ranger to measure velocity understand neither how the sounds are created and
detected nor how the position data is transformed into velocity data. (They do seem to un-
derstand the idea that the sound’s travel time gives a measure of distance—at least qualita-
tively.) On the other hand, I don’t require that they know how their calculator calculates the
sine function before I permit them to use it. When I have them carry out some indirect ac-
tivity, such as producing a spark tape from some motion, making measurements of the posi-
tions, and then calculating and graphing the result, I may be helping them to understand
how to make measurements, but the time delay between the motion itself and the produc-
tion of the graph may be 15 minutes or more. This is far too long for them to buffer or re-
hearse their memory of the motion and make an intuitive connection.
A good example of how this works is the first RTP activity, “Introduction to Motion.”
This is the first lab and begins by having students use a sonic ranger motion detector (see
Figure 8.11) to create position graphs of their own motions.15 They see how the apparatus
works by performing a series of constant-velocity motions and seeing what position graph
appears. I refer to this process as a psychological calibration. They quickly get the idea and
identify some interesting and relevant experimental issues, such as getting out of range of the
beam of sound waves, getting too close (the motion detector only works at distances greater
than about 50 cm), and seeing “bumps” produced by their individual steps. They then make
predictions for a specific motion described in words, do the measurement, and reconcile any
14 Seechapter 2 and the discussion of primitives and facets.

15 The ranger works be emitting clicks from a speaker in the ranger and measuring the time until an echo returns
and is detected by a microphone in the ranger.
Figure 8.11 A sonic ranger motion detector from Vernier software.
discrepancies. Finally, they carry out a variety of position graphs in order to see a broader
range of possibilities.
The experiment then turns to a study of velocity graphs. Again, they begin with a psy-
chological calibration, measuring simple constant-velocity motions, inspecting the graphs, and
comparing them to the position graphs. They then undertake an interesting activity: match-
ing a given velocity graph, the one displayed in Figure 8.12. The graph is cleverly chosen. An
initial velocity of 0.5 m/s away from the detector for 4 seconds produces a displacement of
In this activity, you will try to move to match a velocity—time graph shown on the computer
screen. This is often much harder than matching a position graph as you did in the previous
investigation. Most people find it quite a challenge at first to move so as to match a velocity
graph. In fact, some velocity graphs that can be invented cannot be matched!
1. Open the experiment file called Velocity Match (L1A2-2) to display the velocity—time graph
shown below on the screen.
+1
Velocity (m/s)
−1
0 4 8 12 16 20
Time (s)
Prediction 2-2: Describe in words how you would move so that your velocity matched each part
of this velocity-time graph.
2. Begin graphing, and move so as to eliminate this graph. You may try a number of times. Work
as a team and plan your movements. Get the times right. Get the velocities right. Each person
should take a turn. Draw in your group’s best match on the axes above.
Question 2-4: Describe how you moved to match each part of the graph. Did this agree with
your predictions?
Question 2-5: Is it possible for an object to move so that it produces an absolutely vertical line
on a velocity—time graph? Explain.
Question 2-6: Did you run into the motion detector on your return trip? If so, why did this
happen? How did you solve the problem? Does a velocity graph tell you where to start? Explain.
Figure 8.12 An activity from a RealTime Physics lab. Students are making use of a sonic ranger at-
tached to a computer to display velocity graphs in live time.
2 m. A return velocity of –0.5 m/s for 6 seconds produces a displacement of 3 m. Although

I have not run these laboratories at the University of Maryland, we have adapted them for
the Activity-Based Physics Tutorials. In their prediction, students describe the motion required
in terms of the velocities—speeds and directions—but rarely think about distances. (I’ve never
seen anyone do it.) As a result, they start at the minimum distance from the ranger and be-
gin by walking backward. After having gone back 2 meters, they try to come forward 3 me-
ters and run into the detector. In their attempt to resolve the difficulty (sometimes a care-
fully placed question from the facilitator is needed), they effectively explore the relation
between a velocity graph and the resulting displacements. They are then asked to predict po-
sition graphs given a velocity graph.
The RealTime Physics laboratory continues with an exploration of average velocity and
with fitting the graph with a straight line using computer tools that are provided. Specific
questions try to ensure that the students actually think about the result the computer is giv-
ing and do not simply take it as a given. Finally, the lab ends with a measurement of the ve-
locity of a cart on a track in preparation for the second experiment, which is concerned with
acceleration.
The above example illustrates many of the features common in RTP labs. They often
include:
1. Psychological calibration of the measuring apparatus

2. Qualitative kinesthetic experiments (using one’s own body as the object measured)
3. Predictions
4. Cognitive conflict
5. Representation translation
6. Quantitative measurement
7. Modeling data mathematically
RTP labs are effective in building concepts

The psychological calibration, live-time graphs, and kinesthetic experiments appear to have
a powerful effect on intuition building. In my experience, students who have done these ex-
periments are much more inclined to make physical sense of velocity and acceleration than
students who have done more traditional experiments. Thornton and Sokoloff report data at
their home institutions on subsets of the FMCE16 with RTP laboratories [Thornton 1996].
At Tufts, Thornton tried the RTP laboratories with an off-semester calculus-based class of
about 100 students. He reports that off-semester students tend to be less well prepared than
those who begin physics immediately in the fall. The results are shown in Table 8.3. The
gains were excellent. (The two rows report a cluster of questions probing Newton’s first and
second laws in a natural language [n] environment17 and in a graphical [g] environment—
the velocity graph questions.)
These results are compelling. The fractional gains achieved by the students with RTP
are exceptional: greater than 0.8. Sagredo is skeptical. He points out that there is no direct
16 See the discussion of the FMCE in chapter 5. The FMCE is on the Resource CD associated with this volume.
17 These are the questions on the FMCE involving the sled. See the FMCE on the Resource CD.
TABLE 8.3 Pre- and Post-Results for Results on the FMCE’s Sled [n] and Velocity Graph
Questions [g] in Different Environments.
Tufts Tufts Oregon Oregon Oregon Oregon
RTP RTP NOLAB NOLAB RTP RTP
(pre) (post) g (pre) (post) g (pre) (post) g
NI&II [n] 34% 92% 0.81 16% 22% 0.07 17% 82% 0.78
NI&II [g] 21% 94% 0.92 9% 15% 0.07 8% 83% 0.82
“head-to-head” comparison with a traditional laboratory and that the instruction was carried
out at the primary institution. Moreover, the test used was developed by the researchers who
developed the instruction. Sagredo is concerned that the extra hours of instruction the stu-
dents had in laboratory (compared to those with no laboratory) might have made a big dif-
ference and that the instructors might have “taught to the test.”
On the first issue, Sagredo, I’m not so concerned. It’s my sense that traditional labora-
tories contribute little or nothing to conceptual learning. In fact, instructors in traditional
courses making their best effort rarely produce fractional gains better than 0.2 to 0.35. Gains
of 0.8 or 0.9 suggest that the method is highly effective.
The second item is of more concern. Of course, in a sense we always “teach to the test.”
As discussed in chapter 5, if the test is a good one, it measures what we want the students to
learn. The problem occurs when instructors, knowing the wording of the questions to be used
in an evaluation, focus—perhaps inadvertently—on cues that can lead students to recognize
the physics that needs to be accessed for a particular question.
To deal with this issue, we have to see how well the instructional method “travels.” The
RTP/ILD dissemination project (supported by FIPSE) sponsored the implementation of RTP
Frequency
Gain
0.1 0.2 0.3 0.4 0.5 0.6 0.7
Traditional instruction ILD only

RTP only RTP and ILD
Figure 8.13 Fractional gains on the FMCE in five different colleges and universities that used fully
traditional instruction, or traditional lectures with RTP, or RTP and ILDs (N 1000) [Wittmann 2001].
Frequency
Gain
0.1 0.2 0.3 0.4 0.5 0.6 0.7
Traditional RTP only

RTP and ILD
Figure 8.14 Fractional gains on the ECCE at four different colleges and universities that used fully
traditional instruction, or traditional lectures with RTP, or RTP and ILDs (N 797) [Wittmann 2001].
laboratories and Interactive Lecture Demonstrations (ILDs—see chapter 7) at a number of

different colleges and universities. Preliminary results using the FMCE as a pre- and post-test
show that the RTP Mechanics lab alone produces substantial improvement compared to
schools using traditional laboratories [Wittmann 2001]. When the RTP Mechanics labora-
tories are supported by the use of ILDs in lecture, the results tend to be even better. These
results are displayed in Figure 8.13.
I infer that the RTP laboratories can have a significant effect on student understanding
of basic concepts and on their ability to use a variety of representations in thinking about
these concepts. They tend to spend less time on error analysis than traditional labs, but since
it is my sense that students in traditional labs typically “go through the motions” of error
analysis but gain little real understanding ([Allie 1998] [Sere 1993]), this may be a case of
giving up two birds in the bush in order to get one in the hand.
A preliminary version of the RTP Electricity labs was available for the FIPSE dissemi-
nation project. A preliminary analysis of the first-year implementation of these labs at sec-
ondary institutions using the Electric Circuits Concept Evaluation (ECCE)18 pre and post
shows strong gains with RTP over traditional and even stronger gains when RTP Electricity
labs are supported by the use of ILDs in lecture (see Figure 8.14).
18 The ECCE is on the Resource CD associated with this volume.

CHAPTER 9
Workshop and Studio Methods
I had become thoroughly disillusioned

by the ineffectiveness of the large general lecture courses
of which I had seen so much in Europe and also in Columbia,
and felt that a collegiate course in which laboratory problems
and assigned quiz problems carried the thread of the course
could be made to yield much better training, at least in physics.
I started with the idea of making the whole course self-contained . . .
I abolished the general lectures.
This general method of teaching . . . has been followed
in all the courses with which I have been in any way connected since.
Robert A. Millikan [Millikan 1950]
The Millikan quote in the epigraph shows that dissatisfaction with traditional lectures is
not a new story. Although Millikan’s Autobiography, from which the quote is taken, was
published in 1950, the course he is describing was introduced in the first decade of the
twentieth century. Many physicists, myself included, have the strong intuition that the em-
pirical component in physics is a critical element and one that introductory students of-
ten fail to appreciate. This failure may occur in part because traditional lectures tend to
be a series of didactic statements of “discovered truth” followed by complex mathe-
matical derivations. An occasional ex-post-facto demonstration or laboratory experiment
“to demonstrate the truth of the theoretical result presented in lecture” does little to
help the student understand the fundamental grounding and development of physical ideas
and principles in careful observation.
There are clearly many possible ways of remedying this oversight. Lectures could be-
gin with the phenomena and build up the concepts as part of a need to describe a set
of phenomena. Laws could be built from observed systematics in the behavior of physi-
cal systems. Laboratories could be of the guided discovery type and could introduce the
material before lecture.
But perhaps the most dramatic modification of an introductory physics course is
to adopt Millikan’s method, in which “laboratory problems and assigned quiz problems
170
Physics by Inquiry • 171
carried the thread of the course.” In the modern era, this approach has been developed
under the rubric of workshop or studio courses, courses in which lecture plays a small (or
nonexistent) role. All the class hours are combined into blocks of time in which the stu-
dents work with laboratory equipment for most of the period.
Perhaps the first modern incarnation of this approach is Physics by Inquiry (PbI), a course
for pre- and in-service teachers developed by Lillian McDermott and her collaborators at the
University of Washington over the past 25 years [McDermott 1996]. In this class, there are
no lectures at all. Students work through building the ideas of topics in physics using carefully
guided laboratory manuals and simple equipment. Although PbI is explicitly designed for pre-
service teachers and other nonscience majors, it is deep and rich enough that many of the les-
sons provide valuable ideas for the development of lessons even for calculus-based physics.
The PbI method was adapted for calculus-based physics in the late 1980s by Priscilla
Laws of Dickinson College under the name Workshop Physics (WP). Since problem solving
and developing quantitative experimental skills are goals not shared by the pre-service teacher
class, Laws expanded McDermott’s vision to include substantial components of modern
computer-based laboratory tools, including computer-assisted data acquisition and data ac-
quisition from video. (She and her collaborators developed many of these tools themselves.)
As set up at Dickinson, Workshop Physics runs in classes of 25 to 30 students. This
is possible at a small liberal arts college like Dickinson where few students take introductory
calculus-based physics.1 Research-based institutions with engineering schools might have as
many as 1000 students taking calculus-based physics in any particular term. Two attempts to
bring something like WP to environments with large numbers of students occurred in the 1990s
at Rensselaer Polytechnic Institute (Studio Physics) [Wilson 1992] [Wilson 1994] and North
Carolina State University (SCALE-UP ). The latter is described as a case study in chapter 10.
PHYSICS BY INQUIRY
Environment: Workshop.
Staff: One trained facilitator per 10–15 students.2
Population: Pre- and in-service K-12 teachers; underprepared students; nonscience
majors.
Computers: Limited use.
Other Equipment: An extensive list of traditional laboratory equipment.
Time investment: Large.
Available Materials: A two-volume activity guide [McDermott 1996]. The Washing-
ton group runs a summer workshop to help interested instructors learn the approach.3
1 Though the numbers grew substantially after the introduction of WP requiring the creation of multiple sections.
2 Ina remarkable experiment, the course has been taught with reasonable success using a single experienced instructor
for 70 students [Scherr 2003].
3 A video (Physics by Inquiry: A Video Resource) is available that provides illustrative examples of the materials being
used. Contact the UWPEG for information.

172 • Chapter 9: Workshop and Studio Methods
Figure 9.1 A simple apparatus from Physics by Inquiry.
One of the earlier modern prototypes of a full studio course was Physics by Inquiry (PbI),
developed by Lillian McDermott and her colleagues at the University of Washington
[McDermott 1996]. The course was developed for students studying to be teachers (pre-
service teachers in the American terminology) and is a full guided-discovery laboratory. There
is no lecture; students meet for three laboratory periods of two hours each per week. During
these periods, students work in pairs with simple equipment and are guided to reason through
physical examples with simple apparatus and carefully prepared worksheets. A sample appa-
ratus for the unit on light is shown in Figure 9.1.
In PbI, students learn a few topics deeply

An assumption built into the material is that it is more important for the students to learn a
few topics deeply and to build a sense of how the methods of science lead to “sense-making”
about the physical world than to cover a large number of topics superficially. The materials
emphasize specific concepts and specific elements of scientific reasoning such as control of vari-
ables and the use of multiple representations. The material is structured into independent mod-
ules (see Table 9.1), so a one- or multisemester term can be built by selecting two to three
units per term. This has the advantage that if one permutes the choice of modules in succes-
sive years, in-service teachers can return to take the class in multiple terms without repeating
material.
The worksheets are based on research in student understanding4 and often use the cog-
nitive conflict model in the elicit/confront/resolve form described in the discussion of Tutori-
als in chapter 8. The worksheets guide the students through observing physical phenomena,
constructing hypotheses to explain the phenomena, and the testing of those hypotheses in
new experiments. Trained facilitators (approximately one for every 10 to 15 students) help
students to find their own path to understanding by guiding them with carefully chosen ques-
tions. Specific places are indicated in the lessons called checkouts. Students are instructed to
check their results with a facilitator at this point before going on.
4 Surprisingly,
the Washington group has published very little of their research that has gone into the construction
of Physics by Inquiry. Much of the group’s published work on Tutorials (see references in [McDermott 1999] con-
tained in the Appendix) on qualitative reasoning carries implications for PbI, despite the difference in populations.
TABLE 9.1 Modules in Physics by Inquiry

Volume I Volume II
• Properties of Matter • Electric Circuits
• Heat and Temperature • Electromagnets
• Light and Color • Light and Optics
• Magnets • Kinematics
• Astronomy by Sight: • Astronomy by Sight:
The Sun, Moon, and Stars The Earth and the Solar System
During my sabbatical at the University of Washington (1992–1993), I participated in

facilitating PbI classes. I was particularly impressed by the activity in the Astronomy module
in which students made their own observations of the phase of the Moon and its position
relative to the Sun over the entire term. Near the end of the term, the class’s data were col-
lected and discussed, and a model for how the Moon was lit was developed. Many students
were surprised that they could see the Moon in the daytime, and many believed that the
phases were caused by the Earth’s shadow—a belief they could not sustain in light of the ev-
idence. I myself realized for the first time that I could tell directions from the phase and po-
sition of the Moon, even after sunset.
Students may need help in changing their expectations for PbI

Physics by Inquiry is quite challenging for many students (even physics graduate students),
as the goals, the structure of the learning environment, and the activities expected of the stu-
dent differ dramatically from those they have learned to expect in traditional science classes.
Some students at first resent the idea that they are not being given answers to memorize but
that they have to work them out for themselves and have to understand how the laws and
principles are supported by experiment. Students can exert considerable pressure on an in-
structor to change this. Careful facilitation is needed throughout the course to help students
pay attention to what they are supposed to be doing—thinking, reasoning, and making sense
of what they see in a coherent and consistent fashion. The first few weeks of a PbI class can
be quite tumultuous, but it is worth riding out the storm. The Washington group offers both
extended summer workshops in Seattle and short workshops at meetings of the American
Association of Physics Teachers to help would-be PbI-ers learn the ropes.
Evaluations of PbI show it to be very effective

Although there is not a large body of published literature on the success of PbI, the obser-
vations of a few researchers on secondary implementations of PbI are worth mentioning.
In a recent paper, Lillian McDermott and her colleagues reported on a secondary im-
plementation of PbI for pre-service elementary school teachers at the University of Cyprus
[McDermott 2000].5 They evaluated the performance of students on direct current circuits
5 The classes that used PbI used it in Greek translation.

100%
80%
60%
40%
20%
0%
Present PbI Past PbI Other
Student category
Figure 9.2 Post-test results on the DIRECT concept survey given to students at the University of
Cyprus recently completing PbI, completing PbI in the previous year, and recently completing a more
traditional constructivist physics course for teachers [McDermott 2000].
using the DIRECT conceptual test of understanding of DC circuits. (The DIRECT survey
is on the Resource CD associated with this volume.) Three groups of students were com-
pared: 102 students who had just completed the electric circuits module of the PbI course;
a group of 102 students who completed the module in the previous year; and a group of 101
students who had just completed the topic in a course using constructivist pedagogy but not
using the findings of discipline-based research or the research-redevelopment cycle. (See Fig-
ure 6.1.) The results are shown in Figure 9.2.
Beth Thacker and her colleagues at the Ohio State University compared student success
on a qualitative circuits problem in her secondary implementation of a PbI class with the
same pair of problems given to engineering physics students, students in an honors physics
class, and a traditional physics class for nonscience majors [Thacker 1994]. One problem they
referred to as a synthesis problem. It required only qualitative reasoning. (See Figure 9.3.) A
second problem they referred to as an analysis problem. It required quantitative (algebraic,
not numeric) reasoning. (See Figure 9.4.) The instructor in the engineering class thought that
the problem was exactly appropriate for his students and that they should have little diffi-
culty with it. The instructor in the honors physics class thought the problem was too easy
but was willing to give it as extra credit.
The PbI students scored significantly better than either of the other groups on the syn-
thesis (qualitative) problem and significantly better than the engineers on the analysis prob-
lem. (See Figure 9.5.) Note that an answer was not considered to be correct unless the stu-
dent gave an explanation that included a reason. A restatement of the result (e.g., “Bulb D
is unaffected.”) was not considered sufficient.
A preliminary study of the Ohio State PbI students using the MPEX showed significant
gains on the concept variable [May 2000].
(Synthesis A). All of the bulbs in the figures below have the same
resistance R. If bulb B is removed from the circuit, what happens to the
current through (brightness of) bulb A, bulb D and the battery? Indicate
whether it increases, decreases, or remains the same. Explain your
reasoning.
A
D
B C
(a)
(Synthesis B). A wire is added to the circuit in the figure below. What
happens to the current through (brightness of) bulb A, bulb D and the
battery? Indicate whether it increases, decreases, or remains the same.
Explain your reasoning.
A
D
B C
(b)
Figure 9.3 A qualitative reasoning (synthesis) circuits problem given to test Physics by Inquiry stu-
dents [Thacker 1994].
(Analysis A) What is the total resistance of the network shown in the

figure below? (All of the bulbs have the same resistance.) Show your
work.
(a)
(Analysis B) What is the current through (brightness of) each bulb and
through the battery? (All of the bulbs have resistance R and the current
through the bulb on the right is I as indicated.) Show your work.
(b)
Figure 9.4 A quantitative reasoning (analysis) circuits problem given to test Physics by Inquiry stu-
dents [Thacker 1994].
Synthesis Problem Analysis Problem
Honors
100 100 96%
Inquiry 75%
90 90
Honors 57%
Honors 57%
Percentage completely correct 80 80
Percentage completely correct
Inquiry 54%
70 70
60 60
Engineers 30%
Inquiry 33%
Inquiry 29%
Inquiry 29%
50 50
40 40
Inquiry 17%
Honors 14%
Engineers 8%
Non-science 3%
Non-science 3%
Non-science 3%
Engineers 6%
Non-science 0%
Non-science 0%
Non-science 0%
Engineers 3%
30 30
Engineers 2%
Engineers 2%
Honors 7%
Honors 4%
20 20
10 10
0 0
Whole Part A Part B Whole Part A Part B
problem only only problem only only
Figure 9.5 Results on the electric circuit synthesis and analysis problems given to honors physics, PbI,
engineering physics, and physics for nonscience classes [Thacker 1994].
WORKSHOP PHYSICS
Environment: Workshop.
Population: Introductory calculus-based physics students.
Computers: One for every two students.
Other Equipment: Computer-assisted data acquisition devices (ADCs) and probes,
spreadsheet software, Videopoint™ (video data analysis tools), standard laboratory
equipment.
Time investment: Large.
Available Materials: An activity guide [Laws 1999]. Extensive sets of homework prob-
lems and other resources are available at the WP website: http://physics.dickinson.edu/
A listserve promotes discussions among WP users.
The Workshop Physics (WP) class was developed by Priscilla Laws and her collaborators at
Dickinson College [Laws 1991] [Laws 1999] using the research-redevelopment cycle discussed
in chapter 6. In the mid-1980s, Laws became deeply involved in the use of the computer in
the laboratory, developing laboratory tools for working with Atari computers. In the late
Workshop Physics • 177
1980s, Laws and Ron Thornton of Tufts University, working with a number of fine young
programmers, developed a “stable platform” for microcomputer-based laboratory activities.
The Universal Laboratory Interface box (ULI) is an analog-to-digital converter.6 One end
connects to the computer’s serial port and the other to a “shoebox full” of probes—motion
detectors, force probes, temperature sensors, pressure gauges, voltage probes, and so on. (The
ADCs from Vernier and Pasco are shown in Figure 8.10. The Vernier motion detector is
shown in Figure 8.11.) Software, available for both Wintel and Mac environments, allows
the students to display graphs of any measured variables against any others, to fit the graphs
with various mathematical functions, to read values off the curves, to integrate the curves be-
tween chosen limits, and so on. Spreadsheets (and perhaps symbol manipulators) provide the
students with tools for mathematical modeling of their experimental results.
Students in WP build their concepts using technology

What it is the students actually do in this class is hinted at by the structure of the classroom,
shown in Figure 6.4. The students function in groups as in the inquiry-style classroom, each
pair working with a computer workstation with the computer-assisted data collection struc-
ture and modeling tools described above. Classes are held in three two-hour periods per week.
During these classes, most of the student time is spent with apparatus—making observations
and building mathematical models of their results. The classroom contains a central area for
common demonstrations, and many class periods may include brief lecture segments or whole-
class discussions.
Students are guided through the process of carrying out, making sense of, and model-
ing their experiments with worksheets contained in an Activity Guide [Laws 1999]. In ad-
dition to the Activity Guide, students are assigned reading in a text and homework problems.
Although the homework may include traditional end-of-chapter problems, the WP group
has developed a series of context-rich problems, many of which use video or other computer-
collected data. An example is given in Figure 9.7.
Figure 9.6 Computer-assisted data-acquisition setup showing, from right to left, computer, Vernier
ULI and motion detector, PASCO cart on track.
6 The design of this box was based on previous devices developed by Bob Tinker and his colleagues at TERC.
In the early spring of 1995, the quarterly

Radcliffe College Alumnae magazine featured
a cover story of working women in the
1990s. The unusual cover depicts a young
woman pushing up on a glass ceiling. Assume
that she is not moving. Give three reasons
why the woman’s position in this photo is
impossible. What physical laws or principles
are being violated? How do you think this
photograph might have been made?
Figure 9.7 A sample of a context-rich Workshop Physics problem.
WP is developed through and informed by education research

Although the Dickinson group focuses on development rather than on basic educational re-
search, the development of the Workshop Physics materials relies heavily on published physics
education research and on careful local observations using the research-redevelopment cycle.
An excellent example of how this works is given on the WP web pages. Upon reading the re-
search papers on student difficulty with direct current circuits published by McDermott and
Shaffer [McDermott 1992] [Shaffer 1992], Laws began modifying her WP materials on the
subject. She evaluated students’ conceptual learning on the topic using the ECCE developed
by Sokoloff and Thornton and included on the Resource CD associated with this volume.
She compared her results with those obtained by Sokoloff at the University of Oregon after
students received traditional lectures on the topic. Pre-tests at both Dickinson and Oregon
showed that students entered the class with little knowledge of the subject, missing about
70% of the questions. Lectures helped little, reducing the error rates to about 65%. Students
in Workshop Physics did substantially better, attaining average error rates of as low as 40%.
However, after reading the McDermott-Shaffer papers, Laws redesigned the WP activities.
The results were a substantial improvement, with error rates falling to less than 10%. These
results are shown in Figure 9.8. Similar results are displayed on the WP website http://www.
physics.dickinson.edu for topics in kinematics, dynamics, and thermodynamics.
WP changes the frame in which students work

Implementing Workshop Physics can be a nontrivial activity as the workshop-style class may
violate a number of student expectations. Students who come to a physics class expecting a
Workshop Physics • 179
80 Pre-test
After lectures
After early workshop physics
60 After updated workshop physics
% Error
40
20
0
Electric circuits evaluation results
Figure 9.8 Error rates on the ECCE after traditional lecture (University of Oregon) and after Work-
shop Physics, before and after research-based modifications.
lecture and lots of plug-and-chug homework problems may be dismayed by the amount of
thinking involved. Students who have had high school physics may expect their physics to
be math-dominated rather than experiment-dominated. And students who are unaccustomed
to group work may have trouble interacting appropriately.
Workshop Physics is an attempt to seriously change the framework of learning to have
students focus more strongly on understanding and on the experimental basis of the physics.
Getting students to understand not just the physics but how to make this shift of mental
frame can be difficult. Implementing a course like Workshop Physics effectively requires that
the instructor be sensitive to all these complex issues and be aware of the need to renegoti-
ate the instructor–student social contract.
Traditional
Tutorial/GPS
Workshop Physics
N
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

g
Figure 9.9 Distribution of fractional gains on pre-post FCI and FMCE for traditional, recitation
modifications (Tutorial and CGPS), and Workshop Physics. The histograms for each group are fit with
a normalized Gaussian. The spike at the right corresponds to WP at Dickinson College.
Evaluations of WP show it to be highly effective in building concepts

Jeff Saul and I carried out an independent evaluation of student learning in Workshop Physics
as part of the WP dissemination project (supported by FIPSE) [Saul 1997]. Our study in-
cluded seven colleges and universities implementing Workshop Physics for the first or second
time. Student learning was evaluated with pre-post FCI or FMCE, with common exam ques-
tions, and through interviews with 27 student volunteers at three of the dissemination schools.
Student expectations were measured with the MPEX.
The results from the pre-post FCI/FMCE are schematically shown in Figure 9.9. The
secondary WP implementation averaged fractional gains of 0.41 0.02 (SEM) compared to
0.20 0.03 for the traditional classes and 0.34 0.01 for the recitation modifications. (The
mature primary implementation of WP at Dickinson College typically attains fractional gains
on these tests of 0.74.)
MPEX averages in the traditional classes showed the pre-post deterioration described in
chapter 5. The early secondary implementations showed no significant loss and occasional
small gains on the reality link measure. WP at Dickinson College shows significant gains on
the cognitive cluster of independence-coherence-concepts.
CHAPTER 10
Using the Physics Suite
To suppose that scientific findings decide

the value of educational undertakings is to reverse the real case.
Actual activities in educating test the worth of the results
of scientific results. They may be scientific in some other field,
but not in education until they serve educational purposes,
and whether they really serve or not can be found out only in practice.
John Dewey [Dewey 1929]
We’ve come a long way since I first introduced the idea of the Physics Suite at the be-
ginning of chapter 1. In subsequent chapters, we talked about some of what is known
from research about student thinking and learning, and I described some innovations in
curriculum development based on that research. Some of those innovations belong to
the Physics Suite, while the others can be adapted to work with Suite elements. We are
now ready to revisit the elements of the Suite to consider how you might use them in
your teaching.
The materials of the Physics Suite have been set up so that you can either (1) use
many of them at the same time, or (2) integrate one or more elements with the mate-
rials you are already using. The Suite is not a radical change to the traditional approach
to introductory physics teaching. It is meant to provide elements that are both familiar
and improved as a result of what has been learned from physics education research and
as a result of new developments in educational technology. You can choose to incre-
mentally adopt individual elements of the Suite that are appropriate in your instructional
environment.
In this chapter, I begin with a review of the principles behind the Physics Suite. This
is followed by a brief recap of the elements of the Suite, along with ideas for using the
Physics Suite in different environments. I conclude by presenting four case studies that
give specific examples of how various instructors have adopted and adapted elements of
the Suite in high schools, colleges, and universities.
181
182 • Chapter 10: Using the Physics Suite
THE PRINCIPLES BEHIND THE PHYSICS SUITE

The Physics Suite is predicated on two shifts of perspective from the traditional approach to
teaching, one sociological, one psychological.
First, over the years, physics instructors (myself included) have typically assumed that
they should present the content in a manner that satisfies themselves and that the students
would then take responsibility for doing whatever they needed to do to learn the material.
This results in a filter that passes only those students who come to the class with the drive,
motivation, and understanding of the study skills necessary to succeed in physics on their
own. This turns out to be a small fraction of the students who take physics. With the in-
creasing shift in the emphasis of physics instruction to a service course preparing scientists
and engineers who will not necessarily become physicists, we need to shift our assumptions.
Now we want to see how much help we can offer students who need to know some physics
but who may not know how to learn physics appropriately.
Second, a lot has been learned from educational and psychological research that can help
instructors understand how to help students learn how to learn physics. The critical principle is:1
Principle 1: Individuals build their knowledge by making connections to existing knowl-

edge; they use this knowledge by productively creating a response to the information
they receive.
The implication of this principle is that what matters most in a course is what the stu-
dents actually do. In order to have effective instruction, we, therefore, have to create learn-
ing environments that encourage and enable students to do what they need to do to learn,
even if they don’’t choose (or know how) to do so spontaneously.
This result is often called upon to justify the creation of “hands-on” or “active learning”
environments. Unfortunately, this largely misses the point. Students can actively work with
equipment and still not learn very much physics. (See, for example, the discussion of tradi-
tional laboratories in chapter 9.) What matters is their pattern of thought as they blend hands-
on activities with reflection. We need a more detailed understanding of student learning in
order to be able to design environments that effectively encourage appropriate thought and
reflection. Further complicating the situation is what we need to do to achieve our goals de-
pends on our goals. And these goals depend on a both external and internal factors: the pop-
ulation we are teaching, the course and its nominal purpose, our own individual goals as in-
structors,2 and our model of student thinking and learning.
Our model of thinking and learning is often implicit, tacit, and in contradiction with
both fundamental research in cognitive psychology and the observed behavior of students in
educational situations. In chapter 2, I have put together a “soft paradigm” (a set of guidelines
or heuristics) that can help instructors develop and apply a more sophisticated approach to
thinking and learning. The fundamental ideas are that long-term memory is productive, as-
sociative, and context dependent. Principles 2–5 in chapter 2 (the context, change, individ-
1 Seechapter 2 for details.

2 For example, in teaching our engineering physics class, I very much want my students to learn to understand and
create arguments based on symmetry. Sagredo wants them to appreciate physics as a creative historical evolution of
ideas. Both are noble and justifiable goals but are not required in the classes we teach.
The Elements of the Physics Suite • 183
uality, and social learning principles) help us to understand what features of an environment
might be appropriate to help students create appropriate understandings.
Principle 2: What people construct depends on the context—including their mental states.
Principle 3: It is reasonably easy to learn something that matches or extends an existing
schema, but changing a well-established schema substantially is difficult.
Principle 4: Since each individual constructs his or her own mental structures, different
students have different mental responses and different approaches to learning. Any
population of students will show a significant variation in a large number of cognitive
variables.
Principle 5: For most individuals, learning is most effectively carried out via social
interactions.
This model has a number of implications. Principle 1 implies that it helps if we pay care-
ful attention to what students know and how they use that knowledge in creating both their
correct and incorrect understandings of what we are trying to teach them. This is the resource
component of the model. Principle 2 suggests that it helps if we pay careful attention to when
students access the knowledge we want them to have. This means teaching them how to use
it effectively and to recognize when it is appropriate. This is the linking component of the
model. Principle 3 reminds us of the importance of knowing “what they know that ain’t so”
and provides guidance in building environments that help students get on the right track.
The bridging and cognitive conflict approaches discussed in chapter 2 provide two possible
approaches. Principle 4 reminds us to provide environments appropriate for a variety of stu-
dent styles. This is the diversity component of the model. Finally, Principle 5 suggests we pay
attention to the design of the social environments in which our students learn. This is the
social component of the model.
Chapter 2 also discusses and develops how this model leads us to articulate a particular
collection of explicit goals for our physics instruction. Goals that have been considered in the
construction of the Physics Suite include:
Goal 1: Concepts— Our students should understand what the physics they are learning
is about in terms of a strong base in concepts firmly rooted in the physical world.
Goal 2: Coherence— Our students should link the knowledge they acquire in their physics
class into coherent physical models.
Goal 3: Functionality—Our students should learn both how to use the physics they are
learning and when to use it.
The Physics Suite is designed in accordance with this model of thinking and learning so as
to help teachers create learning environments that function effectively.
THE ELEMENTS OF THE PHYSICS SUITE

Traditional instructional materials are organized around a content list and a textbook. The
Physics Suite is designed to help instructors refocus their courses on learning goals and stu-
dent activities. The text in the Physics Suite is intended to be supportive, but it is just one
component in an array of materials that can help build an effective learning environment.
(See Figure 1.2 for a diagram showing the elements of the Suite.)
I briefly describe each element in turn and show how it fits into the overall picture. Since
the modifications to the narrative—an important Suite element—are not discussed elsewhere
in this volume, I consider these in more detail than the rest.
The Suite’s narrative text: Understanding Physics

The narrative for the Physics Suite, Understanding Physics (Cummings, Laws, Cooney,
and Redish) [Cumming 2003], is based on the sixth edition of the popular Fundamentals of
Physics (Halliday, Resnick, and Walker) [HRW6 2001]. It is being adapted and modified to
an active-learning environment in a number of ways.
1. Modifications to the text are incremental, not radical. The next generation of texts may
integrate activities directly and be presented on-line. Understanding Physics begins with a stan-
dard text and takes a step in that direction. Existing in-text activities (Reading Exercises and
Sample Problems) are enhanced, and links are indicated where connections to activities are
appropriate and relevant. But it still looks enough like a traditional text to be in the “com-
fort zone” of both students and teachers accustomed to traditional texts. The primary active-
learning enhancements come from working with additional Suite elements.
2. The text is modified to take student difficulties into account. At the time of this writing,
thousands of papers have reported on the difficulties students have in learning physics [Pfundt
1994]. These are summarized in the Resource Letter on the Resource CD associated with this
volume [McDermott 1999], and the results are discussed in a number of texts and instruc-
tor’s guides [Arons 1990] [Reif 1995] [Viennot 2001] [Knight 2002]. In Understanding Physics,
issues that are well known to confuse students and cause them difficulty are discussed with
care. Many traditional texts consider these issues trivial and brush them off with a sentence
or use ill-chosen examples that may actually activate classic misconceptions.
3. Topics are introduced by making connections to personal experience whenever possible and
appropriate. We try to follow the principle “idea first, name later” and to motivate a discussion
before it occurs, making contact with the student’s personal experience. This helps build pat-
terns of association between the physics students are learning and the knowledge they already
have, and helps them reinterpret their experiences in a way that is consistent with physical laws.
4. Material is explained in a logical order. We try to follow the “given-new” principle (see
chapter 2) and build on ideas the student can be expected to understand, based on the re-
sources they bring from their everyday experiences. Many texts present results didactically,
starting a discussion by stating a complex result at a point where a student does not have the
resources to understand or interpret it and then explaining it through a complex exposition
taking many pages.
5. Concepts are emphasized. One of the primary goals in our model is that students make
sense of the physics they are learning. This is impossible if they see physics as a set of ab-
struse equations. We therefore stress conceptual and qualitative understanding from the first
and continually make connections between equations and conceptual ideas.
6. Reading exercises and sample (touchstone) problems are limited and carefully chosen. In
an effort to provide examples and items of interest, physics texts often include large numbers
of text boxes, sidebars, and sample problems. This can make the narrative choppy and diffi-
cult for the student to follow. In Understanding Physics, reading exercises are carefully selected
to provide appropriate thinking and reflection activities at the end of a section. These are
often suitable (especially in small classes) as topics for discussion. Sample problems have been
transformed to “touchstones”—carefully chosen examples that illustrate key points to help
students understand how to use the physics within a problem. Sample problems that only il-
lustrated straightforward equation application and manipulation (“plug-and-chug”) have been
removed.
7. Examples and illustrations often use familiar computer tools. Examples in the text have been
expanded and modified to use computer-assisted data acquisition and analysis (CADAA) tools
and collection of data from video. Other elements of the Suite—laboratories, tutorials, inter-
active lecture demonstrations, and Workshop Physics—make heavy use of this technology as
well. This has a number of advantages. It connects the text to the student’s experiences in other
parts of the class; it connects directly to real-world experiences (through video); it uses realistic
rather than idealized data; and it connects the narrative to the more active Suite elements.
8. No chapter summaries are provided. This is a feature, not a bug! We didn’t forget to in-
clude the chapter summaries; rather, we removed them intentionally. Students tend to use
pre-created summaries as a crutch to grab equations for plug-and-chug purposes and as a
shortcut to avoid reading (and trying to make sense of ) the text. Providing an “authority-
validated” summary in the text both robs the students of the opportunity to construct sum-
maries for themselves and sends the covert message to trust authority instead of building their
own judgment. Instructors who feel that summaries are essential (as do I) can assign students
to create summaries as a regular part of their written homework.
9. The order of materials has been modified somewhat to be more pedagogically coherent.
Some of the traditional orderings emphasize the mathematical structure of the material at the
expense of the physics or violate the given-new principle. For example, free fall is often in-
cluded in the kinematics chapter, since constant acceleration problems can be solved alge-
braically. One result of this approach is that students are often confused by gravity, being un-
able to disentangle the idea of the gravitational field near the Earth’s surface (g 9.8 N/kg)
from the gravitational acceleration that results in free fall (ag 9.8 m/s2). We treat free fall
in chapter 3 after a discussion of force. In order to emphasize the centrality of Newtonian
dynamics, Newton’s second law is treated in one dimension immediately after the definitions
of velocity and acceleration. Momentum is treated as a natural extension of Newton’s second
law. The concept of energy is delayed until after the discussion of extended objects.
10. Vector mathematics is handled in a just-in-time fashion. The dynamics of one-dimensional
motion is presented before introducing general two- and three-dimensional vectors. Vectors and
vector products are introduced as they are needed, with the dot product being presented in as-
sociation with the concept of work and the cross product being presented in association with the
concept of torque. One-dimensional motion is presented in the context of one-dimensional vec-
tors, with a notation that is consistent with general vector notation to help alleviate a traditional
confusion students have between scalars and vector components.
Using the Suite in lab: RealTime Physics

RealTime Physics (RTP) is a set of three published laboratory modules covering the topics
Mechanics (12 labs), Heat and Thermodynamics (6 labs), and Electric Circuits (8 labs).3
3A module on Light and Optics is currently under development.

These labs help students build a good understanding of fundamental concepts through use
of a guided inquiry model with cognitive conflict. Experiments rely heavily on computer-
assisted data acquisition to enable students to collect high-quality data quickly and easily. This
allows students to perform many experiments and to focus on phenomena rather than on
data taking. Initial activities with new probes help students “psychologically calibrate” the
probes, that is, convince themselves that they understand what the probes’ responses mean,
even though they may not be clear on how the probes produce their data. Research shows
that these labs can be very effective in helping students build concepts. For a more detailed
discussion, see chapter 8.
Implementing RealTime Physics requires a laboratory setup with computer-assisted data
acquisition equipment for every two or three students.
Using the Suite in lecture: Interactive Lecture Demonstrations

Interactive Lecture Demonstrations (ILDs) help students learn representation-translation skills
and strengthen their conceptual understanding through active engagement in a large lecture
environment. Students receive two copies of a worksheet: one for making predictions and one
for summarizing observations. The instructor goes through a sequence of carefully chosen
demonstrations using computer-assisted data acquisition to display graphs of results on a large
screen in real time. Students are shown the demonstration without data collection. They are
then given the opportunity to make predictions and to discuss their predictions with their
neighbors before the results are collected and displayed. The topics and demonstrations rely
heavily on research that identifies common misconceptions and difficulties. The worksheets
use cognitive conflict and social learning. Research shows that these activities can be very ef-
fective in helping students both learn concepts and understand graphical representations. They
can also be effective in smaller classes. For a more detailed discussion, see chapter 7.
Implementing Interactive Lecture Demonstrations requires only a single computer (for
the lecturer) with computer-assisted data acquisition and a large-screen display. It takes a bit
of practice for a traditional lecturer to develop the interactive style that gets students con-
tributing to the discussion in a way that makes these demonstrations most effective. (See
chapter 7.)
Using the Suite in recitation sections: Tutorials

Tutorials are a curricular environment for delivering active conceptual development in recita-
tion sections. They have a tight, carefully guided group-learning structure similar in feel to the
RealTime Physics labs or the Interactive Lecture Demonstrations. They are based on research
on student difficulties and make frequent use of both cognitive conflict and bridging. An ex-
tensive set of Tutorials has been developed by the University of Washington Physics Educa-
tion Group covering a wide range of topics from kinematics to physical optics [Tutorials 1998].
These Tutorials are designed to be usable in environments without computer tools, so they
make almost no use of computer-assisted data acquisition or video. A supplementary set of tu-
torials using computer technology including computer-assisted data acquisition, video display
and analysis, and simulations, are available as part of the Suite [ABP Tutorials]. Tutorials have
been shown to be effective in improving concept learning compared to classes with traditional
recitations. For a more detailed discussion, see chapter 8.
Implementing the UWPEG Tutorials requires some small items consisting of standard
physics laboratory equipment and inexpensive materials from a hardware store. Implement-
ing the ABP Tutorials requires a computer and data acquisition tools for every three to four
students. Both types of Tutorials require approximately one facilitator per 15 students. These
facilitators need training, both to make sure they understand the physics (which can be quite
subtle and challenging, even for faculty and graduate students in physics) and to help them
learn a “semi-Socratic” approach, in which the instructor guides with a few well-chosen ques-
tions instead of explanations.
Putting it all together: Workshop Physics

Workshop Physics (WP) is the most radical component of the Physics Suite. It presumes a
complete structural change from the traditional lecture/recitation/lab pattern. Typically, the
class is structured into three two-hour laboratory sessions in which the students use sophisti-
cated technology to build their physics knowledge through observation and mathematical mod-
eling. Classes move smoothly back and forth from brief lecture segments, to class discussions,
to full-class demonstrations, to small-group experimenting and modeling. An integrated set of
computer tools are used for data acquisition, video capture and analysis, and graphing and
modeling with spreadsheets.4 Workshop Physics is extremely effective in classes of 30 or fewer,
but it is difficult to deliver to hundreds of students. (See, however, the discussions of the North
Carolina State case study below.) For a more detailed discussion of WP, see chapter 9.
Implementing Workshop Physics requires computer equipment, including a variety of
data acquisition probes and tool software. One facilitator for every 15 students or so is a
must, but if an instructor is present, they can include peer instructors (students who have
successfully completed the course in a previous term). Learning to manage the laboratory lo-
gistics and to help students shift the expectations they might have developed in high school
or other science courses can be a challenge and may take a few semesters before things run
smoothly, but the gains both in learning and in student attitudes can be dramatic.
Homework and exams: Problems and questions

As discussed in chapter 4, the problems students solve, both for homework and on exams,
are a critical part of the activities students carry out to learn physics. The choice of exam
problems is particularly important, since exams send students both overt and covert messages
about what they are supposed to be learning in class (whether we intend to send those mes-
sages or not). Traditional courses often limit homework or exams to questions that have nu-
merical or multiple-choice answers so as to be easy to grade. This has the impact of under-
mining any more sophisticated messages we might send in other parts of the course about
the richness of learning and thinking about physics and the value of learning to make sense
of a physics problem. The Physics Suite includes an enhanced array of problems for home-
work and exams, including estimation problems, open-ended reasoning problems, context-
rich problems, and essay questions.
Implementing more open homework and exam problems requires some structure for
grading. Students need the feedback and motivation that grading provides in order for them
4 These tools are contained on the Resource CD associated with this volume.
to take more complex and open-ended problems with the required degree of seriousness and
reflection. This requires someone—an instructor or assistant—to spend some time evaluat-
ing questions. This can be difficult in large classes, but grading a small number of such ques-
tions (two to three per week, one to two per exam) can have a big impact.
Evaluating instruction: The Action Research Kit

As discussed in chapter 5, over the past 20 years, physics education research has documented
student difficulties in a wide variety of topics in introductory physics. Using these results, re-
searchers have constructed standardized conceptual surveys. Many of the items in these sur-
veys are well designed. They focus on critical issues that are difficult for many students and
they have attractive distractors that correspond to common student misconceptions.5 Because
of the strong context dependence in the response of novice students, these surveys (and es-
pecially a small number of items extracted from a survey) do not necessarily provide a good
measure of an individual student’s knowledge. A broader test with many contexts is required
for that. But these surveys do give some idea of how much a class has learned, especially when
given before and after instruction. More than a dozen surveys are provided on the Resource
CD associated with this volume.
Suite compatible elements

Three non-Suite elements that can be comfortably used in conjunction with other Suite el-
ements materials are Peer Instruction, Just-in-Time Teaching (JiTT), and Cooperative Prob-
lem Solving. These Suite-compatible materials are discussed in detail in chapters 7 and 8.
Peer Instruction
Peer Instruction is a method in which an instructor stops the class every 10 to 15 minutes
to ask a challenging short-answer or multiple-choice question. Usually, the questions are qual-
itative and conceptual and activate a cognitive conflict for a significant number of students.
The students choose an answer for themselves and then discuss it with a neighbor. Next, the
results are collected (by raising hands, holding up cards, or via an electronic student response
system), displayed, and reflected on in a whole-class discussion. Implementing Peer Instruc-
tion only requires a good set of closed-ended questions and problems. Choosing appropriate
and effective problems is not easy. They must reflect a critical conceptual issue, a significant
number of students (20%) must get them wrong, and a significant number of students
(20%) must get them right. As with Interactive Lecture Demonstrations, learning to run
a good Peer Instruction class discussion can take some practice. Mazur’s book on the method
contains a large number of potentially useful problems and helps in getting a good start with
the approach [Mazur 1997]. For a more detailed discussion, see chapter 7.
Just-in-Time Teaching ( JiTT)

JiTT is a method in which students respond to carefully constructed questions (including es-
say and context-rich questions) on-line. The instructor reviews the student answers before
5 Recall from our discussion of the resource component of our learning model that a “misconception” does not nec-
essarily refer to a stored “alternative theory.” It may be produced on the spot by a student using spontaneous asso-
ciations to inappropriate resources or inappropriate mappings of appropriate resources.
Using the Physics Suite in Different Environments • 189
lecture and adapts the lecture to address student difficulties displayed in the answers, some-
times showing (anonymous) quotes for discussion. This method sends the valuable message
that the instructor cares about whether students learn and is responding to them. A signifi-
cant number of appropriately structured problems are contained in the book by Novak, Pat-
terson, Gavrin, and Christian [Novak 1999]. For a more detailed discussion, see chapter 7.
Cooperative Problem Solving

Cooperative Problem Solving is a method for helping students learn to think about complex
physics problems and solve them by working in groups of three in a recitation section or
small class. The method employs heterogeneous grouping of students and assignment of roles,
and it offers the students a structured method to learn to think about how to approach a
complex problem. This method is very effective in helping students both develop good con-
ceptual understanding and learn to solve problems. It sends the valuable message that one
doesn’t have to be able to see how to do a problem immediately in order to solve it, some-
thing many students at the introductory level fail to appreciate. A large number of useful
problems are available on the website of the Minnesota group that developed the method.6
For a more detailed discussion, see chapter 8.
All three of these methods are based on underlying cognitive models and goals similar
to the Physics Suite and coordinate well with it.
USING THE PHYSICS SUITE IN DIFFERENT ENVIRONMENTS

To use parts of the Physics Suite effectively in your classroom requires two elements: a good
match between the Suite elements chosen and the physical classroom conditions, and a good
match between the philosophical orientation of the Suite and the orientation of the instruc-
tors involved.
Some of the Suite elements (RTP, WP, ABP Tutorials) rely heavily on student interactions
with computers and computer-based laboratory equipment. Use of these elements requires ap-
proximately one computer station for every three to four students. These can, of course, be
run in small sections in parallel. One laboratory with 8 to 10 computer stations can easily
serve 400 to 500 students in the course of a week. Some elements of the Suite (RTP, WP, Tu-
torials) require facilitators—one instructor for every 15 students in the classroom. In these en-
vironments, students struggle in small groups with ideas and concepts. They require frequent
(but not too frequent) checking, coaching, and guiding. In principle, one instructor with con-
siderable experience in the methods can handle 30 students (or more), but it is difficult to pull
off. A summary of these physical constraints is given in Table 10.1.
The role of room layout

The room layout plays an important role in using some of the Suite elements effectively. It
is difficult to get students working together effectively in a lecture hall whose chairs are all
oriented in one direction and bolted down. It is difficult to interact effectively with students
working in a computer laboratory in which students sit individually or in pairs at comput-
ers facing in one direction (toward an assumed lecturer) and bolted down in rows. In these
kinds of computer rooms, performing laboratory experiments is nearly impossible. Effective
6 http://www.physics.umn.edu/groups/physed/
TABLE 10.1 Suite Elements Appropriate for a Variety of Environments.

Large Small Facilitator No Computer Computer
Classes Classes Support Facilitator Rich Poor
Element (S/F 50) (S/F 50) (S/F 20) Support (S/C 3) (N 1)
Text (UP) ✓ ✓ ✓ ✓ ✓ ✓
Lab (RTP) ✓ ✓ ✓ ✓
Lecture (ILD) ✓ ✓ ✓ ✓ ✓ ✓
Recitation (UW Tutorials) ✓ ✓ ✓ ✓ ✓
Recitation (ABP Tutorials) ✓ ✓ ✓ ✓
Workshop Physics (WP) ✓ ✓ ✓
room layouts for using various elements of the Suite are discussed in chapters 6 through 9.
These layouts give students the opportunity for face-to-face interaction in small groups.
The role of facilitators

Another important consideration is that the facilitators have the appropriate philosophy and
approach, and that they know how to listen to students and respond appropriately. Despite
the best of intentions, this may not be easy. I had been teaching for 20 years before I real-
ized that when students asked me questions, I was responding as a student rather than as a
teacher. Having been a student for 20 years, having been rewarded for giving good answers
to teachers’ questions, and having been successful at getting those rewards, I had a very strong
tendency to try to give the best answer I could to any question posed. Once I realized (em-
barrassingly late in my teaching career) that the point was not getting the question answered
correctly but getting the student to learn and understand, I shifted my strategy.
Now, instead of answering students’ questions directly, I try to diagnosis their real prob-
lem. What do they know that they can build an understanding on? What are they confused
or wrong about that is going to cause them trouble? As a result, instead of answering a ques-
tion right off, I ask some questions back. Often, I discover that students are trying to hide a
confusion by creating questions that sound as if they know what they are talking about. Help-
ing them to finding resources within themselves that they can bring to bear often makes all
the difference. (“Oh! You mean it’s like . . .”) Even after 10 years of operating in this new
mode, I still detect a strong tendency to want to give “a good answer,” and sometimes, I even
talk myself into believing that for some students, in some situations, it’s appropriate.
Peer and graduate student facilitators may find it particularly hard to be in the right in-
teractive mode. They are still students and tend to easily fall into the mode they use in an-
swering their teachers’ questions. When we first began testing Tutorials at Maryland in the
mid-1990s, my department helped out by letting me handpick some of our best TAs. This
turned out to be a problem. These TAs had developed their reputation by being articulate
explainers. Often in that first semester, I had to pry them out from inside a group of four
students where they had just spent 10 minutes, with pencil in hand, “showing” the students
the answers to all the tutorial questions while the students sat watching, silently.
Finding the right balance of questions and answers, of intervention and “benign neg-
lect,” is difficult. The balance depends on so many things—the particular students involved,
Four Case Studies: Adopting and Adapting Suite Elements • 191
the task, the set of expectations that have been negotiated between student and instructor,
and how tired or frustrated the students are. The key in making the gestalt shift from good
student to effective teacher is learning to listen to the students and to consider them, as well
as the content being discussed.
The large variety of materials offered in the Physics Suite along with the set of Suite-
compatible materials offer instructors a large range of options. Instructors in different envi-
ronments can use the materials in different ways.
FOUR CASE STUDIES: ADOPTING AND ADAPTING SUITE ELEMENTS

Every high school, college, and university physics class is a unique environment. Each has its
own population of students, its own physical environment, its own history of teaching, its
own faculty, and its own relations with other parts of its institution. Any implementation of
new instructional materials must be adapted to each institution’s unique characteristics and
constraints. To illustrate how this plays out in real-world situations, in this section, I present
four case studies of different kinds of institutions that have implemented various elements of
the Physics Suite. The first two cases concern one or a few individuals teaching reasonably
small classes: a public high school and a small liberal arts college. The second two concern
large research universities that teach many students: one without and one with a physics ed-
ucation research group. These stories are based on interviews with some of the faculty in-
volved in implementing the materials, on examination of their materials, and on data from
their websites. I particularly want to thank Maxine Willis, Juliet Brosing, Mary Fehrs, Gary
Gladding, Bob Beichner, and Jeff Saul for discussions.
Using Suite elements at a small institution

Gettysburg High School
Gettysburg High School (GHS) is a medium-sized high school in rural Pennsylvania. It serves
a county that covers 185 square miles and has about 25,000 people. GHS has about 1200
students in four grades. The population draws from a wide demographic, ranging from the
children of professional suburbanites to children who live in rural poverty and who will be
the first generation in their family to attend college.
One of the teachers at Gettysburg, Maxine Willis, has been adapting her class to new
developments in physics instruction over the past 15 years. She now uses many Suite ele-
ments, including Understanding Physics (UP), Workshop Physics (WP), Interactive Lecture
Demonstrations (ILDs), RealTime Physics (RTP), and the WP Tools.
Maxine teaches both a standard physics class (noncalculus) and an AP physics class.7 Typ-
ically, she teaches 40 to 50 students in standard physics divided into two sections and about
30 students in AP physics, again divided into two sections. The classes are therefore reason-
ably small and amenable to highly interactive environments with extended class discussions.
Class periods at GHS are 40 minutes long. Physics is taught in double periods five days
a week to allow for lab work. Once a week, each class also meets in a single period for prob-
lem solving, answering questions, and recitation-like discussions. Since they are using double
periods, they complete a standard one-year high school physics course in one semester.
7This is equivalent to a calculus-based university course in mechanics in preparation for AP Physics C.
Maxine’s classroom is arranged as a Workshop Physics room (see Figure 6.4) and seats
up to 28 students. The room has 14 computers for students, plus one for the instructor. The
instructor’s computer is connected to a flat-plate overhead-projector LCD panel. She is typ-
ically able to get peer instructors for each AP class—students who have previously completed
the class successfully and who get independent study credit for their participation. Accord-
ing to Table 10.1, this makes GHS a small class with facilitator support and a computer-rich
environment. They are therefore able to use all the elements of the Suite.
Maxine has been working with Suite elements in their various development stages since
about 1989. By now she has considerable experience with them and can use them flexibly
and creatively. In the AP class, she uses the text, extensive Workshop Physics activities and
tools, ILDs, and the problem solution book. In the standard class, she uses activities selected
for Workshop Physics, RealTime Physics labs (sometimes substituting the simpler Tools for
Scientific Thinking labs), and ILDs.
In the AP class, two typical days might include a WP activity and a problem-solving ac-
tivity. On a WP day, Maxine may begin by explaining some features of the equipment the
students need to understand to carry out the task, but most of the period is spent with the
students carrying out the activities themselves. Maxine and her peer assistant wander the class-
room, asking and answering questions (and “answers” are often guiding questions). If there
is time left at the end of the period, they may have a reflective discussion of what has been
learned. Otherwise, that discussion takes place at the beginning of the next class.
Typically, WP activities are concept building. Maxine begins a topic with these and doesn’t
turn to serious problem solving until she feels that her students are clear on the concepts. If
WP is not working for them, and in some cases where a WP activity is too complex for high
school or uses too much equipment, she will substitute an ILD, taking a full double period
to complete it.
On a problem-solving day, the students are supposed to have attempted some homework
problems chosen from the text before coming to class.8 They divide into groups of two to
three. Each group is given a piece of whiteboard (2 2) and markers and is assigned a
group number. Maxine then passes out the solution manual, and the students check their an-
swers against the solutions in the manual. While they are doing this, she writes the problem
numbers on the board. When a group decides that they have had difficulty with a problem,
they put their group’s attempt at a solution on the board under the problem number. Both
the instructor and the class can then see the pattern of difficulties. If the entire class has had
difficulty with a problem, Maxine will do a similar example (not the same problem). She then
selects the problems most of the class had difficulty with and has the groups work them out
on their whiteboards.
There are two critical elements in this activity. First, the students have a pattern they
have to follow—they are required to include a diagram and show their line of reasoning. Sec-
ond, the solutions in the manual usually are incomplete. The solutions in the manual pay lit-
tle attention to the problem setup and tend to focus on the algebraic manipulations.9 As a
result, the solutions provide hints but don’t fill in the critical thinking steps; the students
have to do that themselves. By using the hints in the solution book and by working together,
8 Students who fail to do this have their grades for the missed problems reduced.
9 In this context, this is a feature, not a bug!
almost all students are able to work out and understand the solutions to all the problems.
Typically, a class can do three or four problems per time block.
Maxine’s experience with traditional texts in these classes has been poor. Students in the
standard class can’t make much sense of an introductory physics text, and even the AP stu-
dents had considerable difficulty with earlier editions of Halliday, Resnik, and Walker.
Students felt they couldn’t understand it. She helped them by creating reading guides—
questions to help them interpret what they were reading—especially in the first few chapters
when they were getting started. Maxine reports that since adopting the preliminary edition
of Understanding Physics, this problem has gone away. Her AP students are reading the text
carefully and don’t need explicit guidance. (Understanding Physics is not appropriate for her
standard physics students since it uses calculus.)
Maxine reports that the elements of the Physics Suite work well for her and she is satis-
fied that student learning has improved—and not just for memorized facts. When I asked
her what she thought the overall impact of adopting the Suite approach was, she said, “It’s
made me not the center of the classroom. The focus is more on the student as learner. More
of my students are able to think physically. Once they know it, they have a really strong foun-
dation. I feel like I’m giving future scientists their ABCs. I’m not covering a lot of material,
but they’re becoming much more powerful analytically than they were before. In addition,
they’re more confident problem solvers. They go off to competitive colleges and don’t feel
swamped anymore. This is a big improvement. Many of my students used to start out as sci-
ence majors in college and then switch out. That doesn’t happen nearly as often now.”
Pacific University
Pacific University is a private college with professional graduate programs. The professional
programs are mostly in the health sciences (occupational therapy, optometry, physician assis-
tants, and psychology). Many of the undergraduates are interested in biology and in health
science careers. Pacific is located in rural Oregon, in a small town up against the coastal range.
It has about a thousand undergraduates and a thousand students in the professional programs.
The Physics Department is small—four faculty members plus one shared with optome-
try (3.3 FTEs). The Department teaches three introductory physics classes: conceptual physics,
algebra-based physics, and calculus-based physics. Elements of the Physics Suite have been
used in the latter two classes for a number of years.
The algebra-based class has become substantially smaller recently, since the Biology De-
partment no longer requires it.10 The number of students now fluctuates between 20 and 60.
It is taught as a two-semester course meeting six hours/week as three one-hour lectures and
one three-hour lab period. The three-hour period is split between tutorial instruction and
laboratory.
The Pacific Physics Department has been using the RealTime Physics materials in their
laboratories for about eight years. Most of the RTP labs are designed for three-hour blocks,
so they have adapted them to their environment. Many of the RTP labs come in three parts,
each appropriate for a one-hour period. They experimented with a variety of options and
wound up choosing two of the three parts of an RTP lab and splitting it over two weeks. In
10 MaryFehrs reports that according to the biologists the students have to learn so much new biology that there is
no room for courses that are of limited relevance.
each three-hour block, students do a two-hour tutorial and one hour of lab. (They found
that their students were able to handle splitting the lab better than they were able to handle
splitting tutorials.) The RTP labs follow the new mechanics sequence, which does not jibe
with some standard texts. Rather than change the order of reading the text, they keep the
text order and rearrange the order of the RTP labs to match. This seems to work OK.
Some course items (such as circuits) are taught in lab and are only mentioned briefly in
lecture. The instructors at Pacific have chosen not to use the RTP pre-lab assignments since
their students don’t use these materials as a probe of their own thinking, as intended, but
rather look the material up in the text to be sure of getting the right answers. This preempts
the discovery character of the lab learning. The labs work for them without the students hav-
ing completed the pre-lab materials. They do use the lab homeworks but modify them some-
what to fit the language and content of their lectures.
The instructors have struggled somewhat with the tutorial instruction. The UW Tuto-
rials are designed for a calculus-based class and were too sophisticated for their population.
Instructors have been using some problem-solving tutorials developed for the algebra-based
class but are not satisfied with them and so are still hunting for appropriate materials that
emphasize concept building. They do tutorials as a two-hour block in the three-hour period.
Mary Fehrs reports that in lecture she tries to put the material in context and tie it to
concepts, and that she does some problem solving. She also does some ILDs and finds them
very helpful. She reports that in ILDs her students can often be coaxed to make a thought-
ful prediction—which they don’t often do in lab or tutorial. They can be wrong and very
confident about their wrong answers. Sometimes they have to see a result twice before they
believe it. She tried using some JiTT over the web but gave it up because of difficulties man-
aging their computer environment.
Mary’s sense is that the algebra-based students are “good students”—that is, they will do
just about whatever you ask of them. Unfortunately, their mode of successful learning up to
this class has been to memorize and replay, and traditional lecturing plays right into that
mode. The use of RTP and ILDs helps break this pattern and leads to considerable im-
provement. Mary says, “I’ve taught for 30 years. From daily walking and talking you get an
idea of what they’re getting. They get much more this way. They’re really starting to think.”
The class results on FMCE show fractional gains of about 0.5—which is very good compared
to the 0 to 0.3 found in a typical lecture-based environment [Wittmann 2001].
In the calculus-based class, the instructors adopt a full workshop model and use the
Workshop Physics materials. Typically, they have about 20 students and a few peer instruc-
tors (students who have previously taken the course who are paid to facilitate during class
and do grading). The physical setup consists of 1950s-style lab rooms with six long tables,
set up with four students per table. Each table has a computer and an analog-to-digital con-
verter for data acquisition. The students work in groups of twos, but the setup fosters inter-
actions between two pairs. The classes meet for six hours/week in three two-hour blocks.
In this case, they use the Workshop Physics materials as is, without modification. They
haven’t used a textbook (though as of this writing they are planning to try the preliminary
edition of Understanding Physics) and have used WP problems exclusively. They write new
problems for exams but do not feel the need to create additional curricular materials. Mary
Fehrs says, “[The WP materials] make a coherent whole and good conceptual sense as is.”
Some of their exams include components that are laboratory oriented—analysis of data, work
with spreadsheets, etc.
When I asked what problems she had encountered, Mary reported two problems with
WP: the students have difficulty getting the “big picture” working with WP alone, and she
has trouble getting the students to take their predictions seriously and “think hard” about
them. In order to help students develop overviews, she has them write weekly summaries de-
scribing what they have learned during the week. She reports that this exercise helps them
get perspective and organize the material somewhat, but she hopes that having a text will pro-
vide the perspective absent in the hands-on-oriented WP activity guide. She continues to
work on finding ways to help students understand what she wants them to do in the pre-
diction parts of the lesson. The results on pre-post testing for their WP students are very
strong fractional gains—about 0.6 [Wittmann 2001].
Mary is quite satisfied with the use of WP for this class and says she would never go
back to lecturing. Her other colleagues have bought in to the method, and she says they
would expect any new hires to continue using the approach. She likes the fact that WP “im-
mediately tells the students that learning is active, not passive.” She accepts the fact that there
is always a diverse response, with some students loving the approach and some hating it. She
says, “It’s a lot of work and we don’t cover as many topics as we used to. But Workshop
Physics doesn’t let you fool yourself into thinking that your students understand something
that they really don’t. There is constant feedback that reminds you what they haven’t learned
yet. It’s easy to fool yourself when you lecture.”
Using Suite elements at a large institution

In both the case studies discussed above, a small number of instructors (one to three) were
dealing with a reasonably small number of students (100). A significant fraction (about a
third) of the students in the United States taking physics in a service course at the college
level do so in large public research universities. These universities may have between 10,000
and 45,000 students, departments with 15 to 75 faculty members, and graduate students to
serve as TAs. Calculus-based physics may serve as many as 500 to 1000 students in each class
in each term. Managing the laboratories, recitations, homework, and exams for an operation
of this scale can be daunting. Because of the large number of students, large lectures seem
inevitable, and often many different faculty members have responsibility for the same class.
Although departmental committees often choose textbooks and content may be constrained,11
faculty are often given considerable leeway in designing their approach to the class. Labora-
tories may be run independently from the lecture/recitation sections. These strongly held cul-
tural constraints can make implementing lasting reform difficult.
Two large universities that have managed reform even within these constraints are the Uni-
versity of Illinois and North Carolina State University. Both are large engineering schools. The
University of Illinois has adopted and adapted a number of Physics Suite elements within the
context of the traditional large lecture/recitation/laboratory environment and has created its own
11 A common textbook and constraints on content permit students to choose different faculty members’ classes in
different terms in order to handle shifts in the scheduling of their other classes.
web-homework tool. North Carolina State, with the help of an on-site physics education re-
search group, has creatively adapted the workshop approach to a large class environment.
The University of Illinois

At the University of Illinois, Urbana-Champagne (UIUC), in the mid-1990s, the department
head, David Campbell, convinced his colleagues that the results coming out of physics edu-
cation research implied that their large-lecture traditional approach to introductory physics
was not as effective as it could be. The motivations and first steps are described in his article
“Parallel Parking an Aircraft Carrier: Revising the Calculus-Based Introductory Physics Se-
quence at Illinois,” for the Newsletter of the Forum on Education of the APS [Campbell 1997].12
The UIUC is a large state university with a large high-quality, research-oriented physics
department and is one of the premier engineering schools in the country. The Physics De-
partment at UIUC offers three semesters of calculus-based physics and two semesters of
algebra-based physics, teaching all classes every term. A total of about 2500 students register
for these classes each semester. The large number of students requires a large number of fac-
ulty and a large infrastructure, including TAs and lab managers.
Before they reformed the program, the Physics Department at UIUC taught classes in a
traditional fashion with lecture sections of 200 to 300 students for three hours/week and recita-
tion and lab sections of 24 students for three to four hours per week. The lecturer was re-
sponsible for all aspects of lecture, recitation, and homework. The TAs planned their sections
largely on their own and mostly answered questions on problem solving by demonstrating the
solutions themselves at the board. Labs were in the standard “cookbook” model and were the
responsibility of a faculty member who had little or no contact with the lecturers.
The result was that neither faculty nor students were happy. Faculty felt that managing
a large lecture section with associated homework and TAs was a difficult and unrewarding
experience. Students mostly expected to dislike physics and found that the course confirmed
their expectations. In pre-post surveys of student attitudes (see Figure 10.1), more than half
of the students said they considered physics “negative or awful,” with the number increasing
in the end-of-semester survey.
In 1995, the faculty agreed to participate in a major reform of the calculus-based physics
class. Computers were available in labs, and graduate students were available to serve as fa-
cilitators. Available space included traditional large lecture halls with fixed tiered seating, small
recitation classrooms with movable chairs, and traditional laboratory space with long tables.
Funds were made available to provide computers and data acquisition devices for the labo-
ratory. According to Table 10.1, this makes UIUC a large class with facilitator support and
a computer-rich environment. Therefore all of the elements of the Suite can be used with the
exception of Workshop Physics.
Since each of the three classes was taught each semester, the reforms had to be imple-
mented “in flight.” The schedule of implementation is shown in Figure 10.2.13 They decided
12 This article and the other articles in the FED newsletter are available on-line at http://www.aps.org/.
13 The plan—figuring out what to do, preparing materials, and implementing the results—should be contrasted with
the cyclic model displayed in Figure 6.1. The lack of a research-based cycle implies that corrections and updates have
to be handled explicitly in some other way.
THE OLD THE NEW
Student attitudes towards physics 102 (fall99) Student attitudes towards physics 101 (fall99)
35 100
No of students 30
No of students
80
25
20 60
15 40
10
5 20
0 0
Enthusiastic
Positive
Neutral
Negative
Awful
Enthusiastic
Positive
Neutral
Negative
Awful
Before course After course Before course After course
Figure 10.1 Results of pre-post attitude surveys at the University of Illinois—Urbana-Champagne,

before and after curriculum reform. Courtesy of Gary Gladding [Gladding 2001].
to restructure the course to provide more active engagement activities for the students and to
provide a more balanced load for the faculty. A primary design criterion was to produce a
more coherent and integrated course—and one that would be seen as belonging to the de-
partment, not whose individual pieces belonged to individual faculty members or TAs.
The calculus-based course was restructured to include two 75-minute lectures, a two-
hour recitation, and a two-hour laboratory each week. In the algebra-based course, they re-
structured to include two 50-minute lectures, a two-hour recitation, and a three-hour labo-
ratory each week. Lecture classes were increased in size so that one of the faculty members
formerly assigned to lecture could be assigned to manage the recitations and homework. This
resulted in a more balanced teaching load. The decision was made to implement Peer
Instruction in lectures, Tutorials and Cooperative Problem Solving (with home-grown
Fa 95 Sp 96 Fa 96 Sp 97 Fa 97 Sp 98 Fa 98 Sp 99
111 111 114

PHYS 106 Design Refinement/Standardization Production
(8) (4) (6) (4) (4) (8) (4)
112 112 112

(8) (4) (6) (4) (4) (8)
113/4 113/4 113/4

(4) (4) (4) (4) (4)
Figure 10.2 Design and implementation schedule used at the University of Illinois—Urbana-Champagne,
to reform the calculus-based physics course [Gladding 2001].
problems) in recitation, and RealTime Physics laboratories.14 In order to create coherence

and a sense of common ownership, the working team meets regularly to discuss what is hap-
pening in the class. Problems for the cooperative-problem-solving sessions are written in com-
mon, as are exam problems. Exams are multiple-choice and machine-graded, and they are
delivered in the evenings to all sections at the same time outside of class hours.
A new emphasis on concepts was introduced, and a web-based homework system was
developed and maintained by Denny Kane, a full-time staff member. (See [Steltzer 2001].)
The web-delivered homework comes in three different formats, each serving a specific peda-
gogical purpose.
1. Linked quantitative problems around a specific physical situation: Brief hints are avail-
able upon request, and immediate feedback (right or wrong) is given.
2. Delayed feedback homework: Similar in structure to the first, but no feedback is given
until after the grading deadline. These are like on-line quizzes.
3. Interactive examples: A single multistep quantitative problem with extensive help
dialogs.
The UIUC Physics Department made a substantial commitment of both funds and staff
in order to implement the program. Existing positions of computer coordination and lecturer
were repurposed for the new structure, and a new Associate Head position was created to
manage the system.
Both students and faculty are enthusiastic about the results. On pre-post happiness sur-
veys, ratings were dramatically improved. (See Fig. 10.1.) At the end of the first semester,
more than 75% reported that they were either positive or enthusiastic about their physics
class. (This compares with less than 20% before the reform.) Far more TAs made the cam-
pus’s “list of excellent TAs” after reform than did before (77% compared to 19% before re-
forms). Faculty are much more comfortable with the teaching load than previously, and cal-
culus-based physics is no longer considered a “killer” teaching assignment. More details can
be obtained from the course website.15
North Carolina State University

The University of Illinois began their reforms with the premise that changing a large class in-
volving thousands of students and dozens of faculty required maintaining a lecture-based for-
mat, and they adapted many research-based curriculum reforms that fit that model. North
Carolina State University began their reforms with a different idea. The Physics Department
at NC State contains a physics education research group led by Bob Beichner. Bob felt quite
strongly that inquiry-based instructional models such as Workshop Physics could lead to dra-
matic improvements in learning over lecture-based models, so he set about trying to find a
way to implement one.
NC State University, like UIUC, is a large state engineering school with a research-
oriented physics department. The Physics Department has between 50 and 60 faculty members
14 Algebra-based physics labs were adapted from pre-publication versions of RTP, and calculus-based labs were cre-
ated on site using a predict-observe-explain model similar to RTP, ILDs, and Tutorials.
15 http://www.physics.uiuc.edu/education/course_revision.html
and instructs more than 5000 students in physics classes each year.16 The largest class is cal-
culus-based physics, a two-semester course with about 500 to 1000 students in each class in
each semester. The reform of the calculus-based introductory physics class was undertaken in
quite a different way from the UIUC as a result of the presence of a PER group. The proj-
ect was begun in the years 1995–1997 with a small class, observers and interviewers from
the research group, and standardized survey instruments to measure student progress. In the
initial phase of the project, all the students’ classes (physics, calculus, chemistry, and intro-
duction to engineering) were done in a coordinated fashion [Beichner 1999]. In later stages,
the project developed a stand-alone method for physics referred to as Student-Centered Ac-
tivities for Large Enrollment University Physics (SCALE-UP). The SCALE-UP project re-
ceived funding for development and dissemination from U.S. government funding agencies
and is currently being adapted at a number of other universities.17
In the initial stage of the project, the approach planned was described to entering engi-
neering students, and they were asked to volunteer to participate. Approximately 10% of the
students volunteered, and half of those were chosen at random to participate in the experi-
mental class. The other half took the traditional class and were used as a control group.
The class was set up to operate in a workshop/studio mode, and material was adapted
from a wide variety of research-based sources, including Workshop Physics, Physics by In-
quiry, Cooperative Problem Solving, and Peer Instruction. Students were organized into groups
of three heterogeneously and the same groups worked together in all their classes. Roles were
assigned, and students received instruction both on how to work in groups and how to ap-
proach complex problems.
Large numbers of computers with data acquisition and modeling tools (spreadsheets and
Interactive Physics) were available, as were graduate student facilitators. Experimentation with
different layouts led them to select round tables with 9 to 12 students and 3 to 4 laptop com-
puters. Before and after views of the physics classroom are shown in Figure 10.3.
Figure 10.3 Views of the physics classroom at NC State before and after the transformations created
by the SCALE-UP project.
16 Number of students enrolled in physics classes each year including summer sessions.
17 See
the NC State SCALE UP website at http://www2.ncsu.edu/ncsu/pams/physics/Physics_Ed/ for current infor-
mation on the project.
TABLE 10.2 Fraction of Students Who Received Grades of C or Better in

All Their Math, Chemistry, Physics, and Engineering Classes.
1995–1996 1996–1997
N Success Rate N Success Rate
Test class 35 69% 36 78%
Control group 31 52%
Traditional 736 52% 552 50%
Classes are run workshop style, with an intermix of brief lecture elements, discussion,
problem solving, laboratory investigations, and modeling. Laboratory segments are often
brief—10 minutes or so (though they sometimes grow to blocks of as much as an hour or
two)—and in response to questions raised in class discussion. UW Tutorials are used but are
broken up into short discussion segments of 10 to 15 minutes. Since there are no formal lec-
tures, students are responsible for reading materials before each class.
Like the reforms at UIUC, the reforms at NC State made significant use of the web, in
particular, the WebAssign environment, developed and supported at NC State. The web is
used for distribution of materials, maintaining the class schedule, and distributing and col-
lecting homework. WebAssign is used both in and out of class to present questions and prob-
lems to the students. Use is also made of Java applets, particularly the Physlet collection
[Christian 2001].
The classes were initially run with a class size of about 30. They were then increased to 54
and currently run successfully with as many as eight tables and 99 students in a room at once.
The results of the initial attempts showed significant success. The rate of good grades
across the group of classes was much higher in the test group than among traditional stu-
dents (and the control group was close to traditional). (See Table 10.2.)
Student success in physics learning was also better than in the traditional class. The av-
erage score on the TUG-K for the test class was 89% 2%, while for the traditional stu-
dents it was 42% 2% (standard error of the mean). On the FCI pre-post, the test class had
an average fractional gain of 0.42 0.06 and 0.55 0.05 in the two years reported. The
control group only achieved an average fractional gain of 0.21 0.04, comparable to the av-
erage reported for traditional classes [Hake 1992][Redish 1997]. On the shared midsemester
exam, the test class did significantly better than the control group that received traditional
instruction (80% to 68%).18 Pre-post MPEX studies showed no change on most variables (a
good result, considering that almost all classes show a significant loss) and a 1.5 improve-
ment on the coherence variable. On other attitudinal variables, students in the trial class
showed substantial improvement in confidence levels, while students in the traditional class
showed declines (especially those in the control group).19
18 In the following semester, in which all students received traditional instruction in E&M (electricity and magne-
tism), no difference was noted between the groups of students.
19 See the project’s annual reports on the NC State website for more details.
Conclusion • 201
As part of the SCALE-UP project, Beichner and his collaborators are building up a large
collection of adapted and modified materials, including short hands-on activities, interesting
questions to consider, and group-based laboratory exercises that require a lab report. Check
the group’s website for information on availability of these materials.20
CONCLUSION
As these case studies show, there are many paths to reform. The particular path you choose
depends on the resources you have available, your constraints, and above all, the opportuni-
ties offered by your most important resources—the individuals in your department who show
an interest in changing physics instruction at your institution. The Physics Suite offers you
and your colleagues tools to work with in your efforts to improve what your students take
from their physics classes.
20 http://www2.ncsu.edu/ncsu/pams/physics/Physics_Ed/.

Physics Suite PDF

Uploaded by

Copyright:

Available Formats

Physics Suite PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Physics Suite PDF

Uploaded by

Copyright:

Available Formats

Preface P-1 8/7/02:7:11 PM

of Maryland in 1970. I have been on the Maryland faculty ever since,

Teaching Physics with the Physics Suite E. F. Redish

Teaching Physics with the Physics Suite E. F. Redish

2. Modular reasoning structures:

Teaching Physics with the Physics Suit E. F. Redish

Getting Feedback from your Students Models of the Classroom

Teaching Physics with the Physics Suit E. F. Redish

Chapter 10 Using The Physics Suite

Teaching Physics with the Physics Suit E. F. Redish

Teaching Physics with the Physics Suit E. F. Redish

Introduction and Motivation

The Master doesn’t talk, he acts.

TYPICAL MATERIALS FOR A PHYSICS CLASS

Summary book Overheads Simulations Solution book

Lots of unrelated pieces

Figure 1.1 Typical layout of materials associated with a physics course.

A NEW ALTERNATIVE: THE PHYSICS SUITE

Figure 1.2 Elements of The Physics Suite

• The students we are teaching have changed.

I organize my discussion of these points around two questions:

1. Who are we teaching and why?

Who are we teaching and why?

The growth of other sciences

The goals of physics for all

Are we already achieving these goals?

Figuring out what doesn’t work and what we can do about it

WHY PHYSICS EDUCATION RESEARCH?3

3 Much of this section is based on my Millikan lecture [Redish 1999].

Knowledge as a community map

ean plane. See, for example, [Flanders 1963].

An emergent phenomenon— Community

Building the community map for education

The impact on teaching of research on teaching and learning

Even good students get the physics blues

1. In the circuit at the right, find the current 8V

2. In the circuit at the left, explain what will

I wouldn’t have believed it if I hadn’t seen it

13 Forexample, [Halloun 1985b].

A bar magnet is hung from a string through its center.

WHAT THIS BOOK IS ABOUT

Cognitive Principles and

He who loves practice without theory

1 This chapter is based in part on the paper [Redish 1994].

THE COGNITIVE MODEL

• Working memory is limited.

ing [RedishJ 1993].

Figure 2.1 A picture of an animal [Frisby 1980].

the shadows under a leafy tree.

Long-term memory is context dependent. By context-dependent, I mean that the cognitive

9 Adapted from [Wason 1966] and [Dennett 1995].

Figure 2.4 An example illustrating linked context-dependent schema.

The key to understanding student reasoning is understanding the patterns of association

This component of context is an example of context affecting activation responses.

Cognitive resources for learning

priate) items in a student’s schema.

1. Robust reasoning structures: Common naïve conceptions

2. Modular reasoning structures: Primitives and facets

16 This argument is made in a slightly different context in [KuhnT 1970].