By Carol A.

Twigg

IMPROVING LEARNING AND REDUCING COSTS:

NEW MODELS FOR

ONLINE
LEARNING very college and university in the United States is discover-

E
ing exciting new ways of using information technology to
enhance the process of teaching and learning and to extend
access to new populations of students. For most institutions,
however, new technologies represent a black hole of addi-
tional expense. Most campuses have simply bolted new technologies
onto a fixed plant, a fixed faculty, and a fixed notion of classroom in-
struction. Under these circumstances, technology becomes part of the
problem of rising costs rather than part of the solution. In addition,
comparative research studies show that rather than improving quality,
most technology-based courses produce learning outcomes that are
simply “as good as” their traditional counterparts—in what is often re-
ferred to as the “no significant difference” phenomenon.1 By and
large, colleges and universities have not yet begun to realize the prom-
ise of technology to improve the quality of student learning and re-
duce the costs of instruction.
Carol A. Twigg is Executive Director of the Center for Academic Transformation at Rensselaer
Polytechnic Institute. The Center’s mission is to serve as a source of expertise and support for those
in higher education who wish to take advantage of the capabilities of information technology to
transform their academic practices.

28 EDUCAUSE r e v i e w
September/October 2003 © 2003 Carol A. Twigg Illustration by Steve McCracken, © 2003 September/October 2003
EDUCAUSE r e v i e w 29
 Preliminary results show that all thirty institutions reduced costs by about
40 percent on average, with a range of 20 percent to 84 percent.

Supported by an $8.8 million The course-redesign projects which are frequently in modular for-
grant from the Pew Charitable focus on large-enrollment, in- mat, according to scheduled mile-
Trusts, the Program in troductory courses in multi- stones for completion.
Course Redesign (http:// ple disciplines, including 5. On-demand help. An expanded support
w w w. c e n t e r. r p i . e d u / the humanities (6), quan- system enables students to receive as-
PewGrant.html) was cre- titative subjects (13), social sistance from a variety of different peo-
ated in April 1999 to ad- sciences (6), and natural ple. Helping students feel that they are
dress these issues. Man- sciences (5). What do these a part of a learning community is criti-
aged by the Center for projects have in common? cal to persistence, learning, and satis-
Academic Transformation To one degree or another, all faction. Many projects replace lecture
(http://www.center.rpi.edu/) at thirty projects share the following time with individual and small-group
Rensselaer Polytechnic Institute, the Pro- six characteristics: activities that take place either in com-
gram is supporting colleges and universi- puter labs—staffed by faculty, graduate
ties in their efforts to redesign instruction 1. Whole course redesign. In each case, the teaching assistants (GTAs), and/or peer
using technology to achieve quality en- whole course—rather than a single class tutors—or online, enabling students to
hancements as well as cost savings. Se- or section—is the target of redesign. have more one-on-one assistance.
lected from hundreds of applicants in a Faculty begin the design process by an- 6. Alternative staffing. By constructing
national competition, thirty institutions alyzing the amount of time that each support systems consisting of various
received a grant of $200,000 each, with person involved in the course spends kinds of instructional personnel, the
the grants awarded in three rounds of ten. on each kind of activity, a process that projects apply the right level of
The thirty institutions include research often reveals duplication of effort human intervention to particular stu-
universities, comprehensive universities, among faculty members. By sharing re- dent problems. Not all tasks associated
private colleges, and community colleges sponsibility for both course develop- with a course require highly trained,
in all regions of the United States. ment and course delivery, faculty save expert faculty. By replacing expensive
The Center has required each institu- substantial amounts of time while labor (faculty and graduate students)
tion to conduct a rigorous evaluation fo- achieving greater course consistency. with relatively inexpensive labor (un-
cused on learning outcomes as measured 2. Active learning. All of the redesign proj- dergraduate peer mentors and course
by student performance and achieve- ects make the teaching-learning enter- assistants) where appropriate, the
ment. National experts have provided prise significantly more active and projects increase the person-hours
consultation and oversight regarding the learner-centered. Lectures are re- devoted to the course and free faculty
assessment of learning outcomes to en- placed with a variety of learning re- to concentrate on academic rather
sure that the results are reliable and valid. sources that move students from a than logistical tasks.
To date, results show improved student passive, note-taking role to an active,
learning in twenty of the thirty projects, learning orientation. As one math pro- Although all thirty projects have these
with the remaining ten showing no signif- fessor put it, “Students learn math by characteristics in common, each has cho-
icant difference. Each institution has doing math, not by listening to some- sen a design model that implements the
also been required to develop a detailed one talk about doing math.” characteristics in a way that varies accord-
cost analysis of both the traditional and 3. Computer-based learning resources. In- ing to the discipline involved, the partic-
the redesigned course formats, using a structional software and other Web- ular student audience, and the prefer-
spreadsheet-based course-planning tool based learning resources assume an ences of faculty. After examining the
(http://www.center.rpi.edu/PewGrant/ important role in engaging students similarities and differences in how these
Tool.html) developed by the Center. Pre- with course content. Resources in- common characteristics are arrayed in the
liminary results show that all thirty insti- clude tutorials, exercises, and low- various projects, the Program has been
tutions reduced costs by about 40 percent stakes quizzes that provide frequent able to identify five distinct course-
on average, with a range of 20 percent to practice, feedback, and reinforcement redesign models: supplemental, replace-
84 percent. Other outcomes include in- of course concepts. ment, emporium, fully online, and
creased course-completion rates, im- 4. Mastery learning. The redesign projects buffet. A key differentiator among them
proved retention, better student attitudes add greater flexibility for when stu- is where each model lies on the contin-
toward the subject matter, and increased dents can engage with a course, but uum from fully face-to-face to fully on-
student satisfaction with the mode of in- the redesigned courses are not self- line interactions with students.
struction. Collectively, the thirty re- paced. Rather than depending on
designed courses affect more than 50,000 class meetings, student pacing and The Supplemental Model
students nationwide and produce a sav- progress are organized by the need to The supplemental model retains the
ings of $3.6 million each year. master specific learning objectives, basic structure of the traditional course,

30 EDUCAUSE r e v i e w
September/October 2003

 The key characteristic of the replacement model is a reduction in class-
meeting time, replacing face-to-face time with online, interactive learning
activities for students.
particularly the number of class of skills in statistical inference— erates the discussions and draws out key
meetings. Some of the sup- in effect, providing a per- issues to reinforce specific ideas or reveal
plemental redesigns simply sonal tutor for each misconceptions. Redesigning in-class ac-
add technology-based, student. After using tivities has encouraged students to come
out-of-class activities to SmartLab, students in- to class; in turn, this increased attendance
encourage greater stu- creased their scores on a has had a positive effect on student learn-
dent engagement with test of skills and concepts ing. At UMass, attendance in the tradi-
course content. Others by 3.65 out of 16 items, for tional format averaged 67 percent; in the
change what goes on in the a 22.8 percent increase, a redesigned course, attendance averaged
class meetings as well as significant improvement, t 90 percent, which correlated significantly
adding out-of-class activities. (19)= 5.877, p< .001. In addition, to performance on exams. In addition,
The redesign of general psychology at SmartLab helped students achieve a level exams no longer emphasize recall of fac-
the University of New Mexico (UNM) of statistical literacy not deemed possible tual material or definitions of terms; 67
(http://www.center.rpi.edu/PewGrant/ in the course before its redesign. percent of the questions now require rea-
RD3%20Award/UNM.html) and the re- The redesign of introductory biology soning or problem-solving skills, com-
design of introductory statistics at at the University of Massachusetts– pared with 21 percent previously.
Carnegie Mellon University (http://www Amherst (http://www.center.rpi.edu/ At UC, the entire introductory astron-
.center.rpi.edu/PewGrant/RD2%20Award/ PewGrant/RD2%20Award/UMA.html) omy class (approximately 200 students)
CMU.html) exemplify the first version of and the redesign of introductory astron- meets twice a week. At the first meeting,
the supplemental model of redesign. omy at the University of Colorado – the instructor provides an overview of
Each institution kept the lecture portion Boulder(UC) (http://www.center.rpi the week’s activities. About a dozen dis-
of the course intact, including the num- .edu/PewGrant/RD1Award/UCB.html) cussion questions are posted on the Web;
ber of class meetings, but supplemented exemplify the second version of the sup- these range from factual questions testing
lectures and textbooks with a variety of plemental model of redesign, changing basic knowledge, to complex questions
computer-based activities. in-class activities as well as adding out-of- requiring students to draw conclusions,
At UNM, students receive credit for class activities. The goal is to create an ac- to questions intended to elicit contro-
completing three online mastery quizzes tive learning environment within a large versy. Midweek, students meet for one
each week. Students are encouraged to lecture hall setting supplemented by a va- hour in small learning teams of 10 to 15
take the quizzes as many times as needed riety of out-of-class activities that ensure students (supervised by undergraduate
until they attain a perfect score. For all students are prepared when they come to learning assistants) to prepare answers
quizzes, only the highest scores count. class. collaboratively and to carry out inquiry-
The more time students spend taking Before class, UMass students review based team projects. Teams are sup-
quizzes and the higher their scores, the learning objectives, key concepts, and ported by software that allows them to
better they perform on in-class exams. A supplemental materials posted on the collaborate synchronously or asynchro-
two-disc CD-ROM, which contains inter- class Web site. To assess their preparation nously. All teams post written answers to
active activities, simulations, and movies, for class, students then complete online all questions, and every team member
is used to review and augment text mate- quizzes, which provide immediate feed- must sign up as a designated answerer for
rial. At UNM, the drop-withdrawal- back to students and data for instructors one or two questions.
failure (DWF) rate in the course has fallen to assess students’ knowledge levels. In- At the next class meeting, the instruc-
from 42 percent in the traditional format structors are able to reduce class time tor leads a discussion session in which he
to 18 percent in the redesign, and the spent on topics that the students clearly directs questions not to individual stu-
number of students who received a C or understand, increase time spent on prob- dents but to the learning teams. Before
higher has risen from 60 percent to 76.5 lem areas, and target individual students the meeting, the instructor uses software
percent. for remedial help. During class, UMass to review all the posted written answers to
Carnegie Mellon has redesigned the uses ClassTalk, a commercially available, a given question. If all the teams have cor-
laboratory portion of its statistics course interactive technology that compiles and rectly answered a given question, the in-
while leaving the lecture portion intact. displays students’ responses to problem- structor skips that question. Instead, he
The redesign uses SmartLab, an auto- solving activities. Class time is divided devotes the discussion time to questions
mated, intelligent tutoring system that into ten- to fifteen-minute lecture seg- with dissonant answers among teams. Pe-
monitors students’ work as they go ments followed by sessions in which stu- riodically, the instructor poses a related
through lab exercises. SmartLab provides dents work in small groups applying con- question and gives the class time for each
them with feedback when they pursue an cepts to solve problems posed by the team to formulate an answer. The discus-
unproductive path and closely tracks and instructor. Group responses are reported sion sessions both reinforce what stu-
assesses individual students’ acquisition through ClassTalk. The instructor mod- dents have learned and clear up miscon-

32 EDUCAUSE r e v i e w
September/October 2003

ceptions. Rather than emphasizing stu- cussion of the results, and 10 percent for homework assignments, and using Web-
dents’ mastery of facts, the redesign is online quizzes on concepts related to the based resources. Students then take the
teaching students to develop their under- activities. Students are regularly tested on tests individually. Immediately following
standing of the scientific process through assigned readings and homework using the individual effort, the students take
written and verbal communication and to Readiness Assessment Tests (RATs), short the same test in groups of four. In addi-
draw conclusions from collaborative quizzes that probe students’ conceptual tion to motivating students to keep on top
inquiry-based activities. understanding. Constituting 30 percent of the course material, RATs have proven
of the students’ grades, RATs are given to be very effective in detecting areas in
The Replacement Model five to seven times during the course. Stu- which students are not grasping the con-
The key characteristic of the replacement dents prepare to take the RATs outside of cepts, enabling faculty to take corrective
model is a reduction in class-meeting class by reading the textbook, completing actions in a timely manner.
time, replacing (rather than supplement-
ing) face-to-face time with online, inter-
active learning activities for students. The
assumption is that certain activities can be
better accomplished online, either indi-
vidually or in small groups, than in a class.
In some cases, out-of-class activities take
place in computer labs; in others, they
occur online so that students can partici-
pate anytime, anywhere. One version of
the replacement model replaces some
class meetings with online activities while
keeping in-class activities more or less the
same. Others replace some class meetings
with online activities and also make sig-
nificant changes in what goes on in the re-
maining class meetings. Rather than as-
suming that face-to-face meetings are the
best setting for student learning, these
projects have thought about why (and
how often) classes need to meet in real
time and the content of that meeting in re-
lation to the desired learning outcomes.
The redesign of introductory statistics
at Pennsylvania State University (Penn
State) (http://www.center.rpi.edu/
PewGrant/RD1Award/PSU.html) and the
redesign of general chemistry at the Uni-
versity of Wisconsin at Madison (UW)
(http://www.center.rpi.edu/PewGrant/
RD1Award/UWM.html) exemplify the
first version of the replacement model of
redesign. Each institution has replaced a
portion of its class meetings with online
activities while keeping the lecture for-
mat in the remaining class meetings.
Penn State’s course redesign involves
reducing lectures from three to one per
week and changing two traditional recita-
tion sections to two computer-studio
labs. In the computer-studio labs, stu-
dents work individually and collabora-
tively on prepared activities. Approxi-
mately 30 percent of the lab time is used
for elaboration of concepts, 60 percent
for computer-related work and class dis-

September/October 2003
EDUCAUSE r e v i e w 33

 The emporium model eliminates all class meetings and replaces them
with a learning resource center featuring online materials and on-demand
personalized assistance.
At UW, Web-based resources meeting times from three to two activities that foster collaboration, profi-
have replaced one of two lec- per week and moved those ciency, and higher levels of thinking.
tures and one of two discus- course aspects that can be
sion sessions per week; the better accomplished The Emporium Model
rest of the course remains using technology to an The emporium model was first devel-
unchanged. Building on o n l i n e e n v i ro n m e n t . oped at Virginia Tech (VT) (http://www
substantial experience in UTK online activities in- .center.rpi.edu/PewGrant/RD1Award/
using and developing clude grammar, vocabu- VA.html). The model is based on the core
interactive materials, UW lary, and listening exercises; idea that the best time to learn mathemat-
has developed thirty-seven PSU’s include testing, writing, ics is when the student wants to do so
Web-based instructional mod- and grammar instruction as well rather than when the instructor wants to
ules in chemistry. Each module leads a as small-group activities focused on oral teach. The redesign model allows stu-
student through a topic in six to ten inter- communication. Students receive imme- dents to choose when to access course
active pages. When the student has com- diate feedback and detailed explanations materials, what types of learning materi-
pleted the tutorial, a debriefing section in response to their online work, and als to use depending on their needs, and
presents a series of questions that test class time is freed for interactive and col- how quickly to work through the
whether the student has mastered the laborative learning experiences. Online materials with the support of sophisti-
content of that module. Students particu- grading has given the instructors more cated instructional software and one-on-
larly like the ability to link directly from a time to prepare their classes and to focus one on-site help. Following the successes
problem they have difficulty with to a tu- on meaningful communicative and col- achieved at VT, the University of Alabama
torial that helps them learn the concepts laborative tasks in class. By making these (http://www.center.rpi.edu/PewGrant/
needed to solve the problem. To help changes, both universities have been able RD2%20Award/UA.html) and the Univer-
students structure their studying, the to increase the time that students spend sity of Idaho–Moscow (http://www.center
chemistry team has also developed 417 in oral communication. Furthermore, .rpi.edu/PewGrant/RD2%20Award/UI
homework question sets that include di- they have been able to increase the num- .html) have replicated the emporium
agnostic feedback pointing out why each ber of students who can be served with model with student bodies that are less
incorrect response is not appropriate. the same personnel resources. prepared to study mathematics.
The redesigns of introductory Spanish Like most other colleges, TCC has tra- The emporium model eliminates all
at the University of Tennessee–Knoxville ditionally taught writing in small sections class meetings and replaces them with a
(UTK) (http://www.center.rpi.edu/ (approximately 30 students). Consider- learning resource center featuring online
PewGrant/RD2%20Award/UTK.html) able class time was spent reviewing and materials and on-demand personalized
and at Portland State University (PSU) reteaching basic skills, thus reducing the assistance. The model requires a signifi-
(http://www.center.rpi.edu/PewGrant/ amount of time during which students cant commitment of space and equip-
RD3%20Award/PoSU.html) and the re- could engage in the writing process. By ment. VT’s Math Emporium holds 500
design of college composition at Tallahas- shifting many basic instructional activi- workstations as well as other specialized
see Community College (TCC) (http:// ties that can be readily individualized to spaces and equipment. The University of
w w w. c e n t e r . r p i . e d u / Pe w G r a n t / technology, TCC’s redesign enables stu- Alabama’s Mathematics Technology
RD3%20Award/TCC.html) exemplify the dents and faculty to focus on the writing Learning Center (MTLC) contains 240
second type of replacement model, in process in the classroom portion that re- computers plus rooms for individual tu-
which some classes are replaced with on- mains. TCC uses technology to provide torial activities. The University of Idaho’s
line activities and the remaining classes various resources: diagnostic assess- version is called Polya and contains 72
are changed. Each institution has re- ments resulting in individualized learn- computers, in pods of four, designed for
designed its entire course by shifting ing plans; interactive tutorials in gram- as many as three students to work to-
many instructional activities to the tech- mar, mechanics, reading comprehension, gether at a single monitor. Moving away
nology while using the classroom portion and basic research skills; online tutorials from the three-contact-hours-per-week
of the course to focus on those activities for feedback on written assignments; norm, the emporium model significantly
that require face-to-face interaction. follow-up assessments; and discussion expands the amount of instructional as-
The most significant academic prob- boards to facilitate the development of sistance available to students: VT’s Math
lem in traditional Spanish courses is that learning communities. These activities Emporium is open 24/7; Alabama’s
about 85 percent of in-class time is spent take place in two labs per week, and the MTLC is open 71 hours per week; Idaho’s
explaining and practicing grammar and resources are accessible to students at any Polya center is open 86 hours per week.
vocabulary instead of practicing the ex- time. In the one class meeting that re- Multiple sections of a course are com-
pressive skills of speaking and writing. mains, students work individually or in bined into one large course structure, re-
Both UTK and PSU have reduced class- small groups on a wide range of writing placing duplicative lectures, homework,

34 EDUCAUSE r e v i e w
September/October 2003

and tests with collaboratively developed port. It soon became apparent that the sonally answering every inquiry, com-
online materials. Virginia Tech has com- undergraduate students were as effective ment, or discussion. As a result, faculty
bined 38 linear algebra sections of ap- as the graduate students in providing tu- members often spend more time teach-
proximately 40 students each into one torial support, thus eliminating the need ing online and interacting with students
1,500-student section; the University of for graduate students. Based on student- than is the case in classroom teaching.
Alabama has combined 44 intermediate use data collected during the first semes- Very few of the courses involved in the
algebra sections of approximately 35 stu- ter of operation, Alabama also reduced Program in Course Redesign are fully on-
dents each into one 1,500-student sec- the number of instructors and under- line, and those that are do not follow the
tion. The University of Idaho has moved graduate tutors assigned to the MTLC by labor-intensive model used by most on-
two precalculus courses, previously or- matching staffing levels to student-use line programs. Instead they adopt many
ganized in 60 sections of approximately trends. of the design principles used by the sup-
40 students each, into its Polya learning As in the supplemental and replace- plemental, replacement, and emporium
center, treating each course as a coherent ment course-redesign models, there are models described above. Rio Salado Col-
entity. Each university, by teaching multi- two types of emporiums. Virginia Tech lege’s redesign of four precalculus math-
ple math courses in its facility, can share follows an open-attendance model, ematics courses and the University of
instructional person-power among whereas Alabama and Idaho have added Southern Mississippi’s redesign of its
courses, significantly reducing the cost of mandatory attendance and required world literature course exemplify the
teaching these additional courses. group meetings to ensure that students model of the fully online course.
The emporium model is heavily spend sufficient time on task. Alabama Rio Salado’s redesign (http://www
dependent on instructional software, in- requires students to spend a minimum of .center.rpi.edu/PewGrant/RD1Award/rio
cluding interactive tutorials, computa- 3.5 hours per week in the MTLC and to at- .html) is based on using Academic Sys-
tional exercises, electronic hyper- tend a thirty-minute group session each tems’ mathematics software and the addi-
textbooks, practice exercises, solutions to week. This session focuses on students’ tion of a nonacademic course assistant.
frequently asked questions, and online problems and allows instructors to follow The Academic Systems software presents
quizzes. Modularized online tutorials up in areas where testing has defined the content of the course so well that in-
present course content with links to a weaknesses. Idaho students are assigned structors did not need to spend time de-
variety of additional learning tools: to focus groups, of 40 to 50 students each, livering content. The addition of a course
streaming-video lectures, lecture notes, according to their majors so that particu- assistant to address non-math-related
and exercises. Navigation is interactive; lar applications can be emphasized. questions (which constituted 90 percent
students can choose to see additional Groups meet once a week to coordinate of all interactions with students!) and to
explanation and examples along the way. activities and discuss experiences and ex- monitor students’ progress frees the in-
Online weekly practice quizzes replace pectations. Both universities believe that structor to concentrate on academic
weekly homework grading. With the the group activities help build commu- rather than logistical interactions with
development of a server-based testing nity among students and between stu- students. As a result, one instructor is
system, large databases of questions are dents and instructors. able to teach 100 students concurrently
easily generated, and grading and record- enrolled in any of four math courses. Be-
keeping are automatic. The Fully Online Model fore the redesign, the instructor typically
Each emporium is staffed by a combi- On most campuses, the job of a faculty taught 35 students in one section.
nation of faculty, GTAs, and peer tutors. member is seen as monolithic: to perform Rather than relying on individual fac-
Instead of spending time preparing lec- a collection of tasks that are, with few ex- ulty members in small sections to provide
tures or grading homework and tests, in- ceptions, carried out alone. American feedback to students, Rio Salado takes ad-
structors and others devote time to re- higher education remains what Bill vantage of the Academic Systems soft-
sponding directly to each student’s Massy and Bob Zemsky have called a ware’s large bank of problems and an-
specific, immediate needs. Emporium “handicraft” industry in which the vast swers for each topic to increase the
helpers do not answer students’ ques- majority of courses are developed and de- amount and frequency of feedback to
tions but rather direct students to re- livered as “one-offs” by individual profes- students. All assignments are completed
sources from which they can learn. By sors.2 In most colleges and universities, within the context of the software and are
creating a kind of triage response team, this repetitive, labor-intensive approach graded on the spot. Because of this imme-
the emporium model increases the num- has been transferred to online education diate feedback, students know which
ber of contact hours for students while it as well. Individual faculty members de- course aspects they have not mastered
greatly decreases the cost per hour for sign and deliver multiple course sections, and are able to take appropriate correc-
that contact. Staffing adjustments can be each of which is relatively small in size. tive actions. The software enables each
made based on real use. For example, Al- Web-based materials are used largely as student to work as long as needed on any
abama’s initial plan was to staff the MTLC supplemental resources rather than as particular topic. Students can take the
primarily with instructors and to use substitutes for direct instruction. This end-of-module quizzes as soon as they
graduate students and upper-level, un- model assumes that the instructor must are ready, moving quickly or slowly
dergraduate students for tutorial sup- be responsible for all interactions, per- through the material depending on their

September/October 2003
EDUCAUSE r e v i e w 35

 Consistent content coverage means that all students have the same
kinds of learning experiences, resulting in significant improvements in
course coherence and quality control.
comprehension and past experi- and quality control. Treating the amount of interaction that they require
ence or education. The soft- whole course as one section with faculty, staff, and one another. At the
ware also provides a built-in also eliminates duplication British Open University, for example, ap-
tracking system that allows of effort on the part of in- proximately one-third of the students
the instructor and the structors; faculty in- never interact with other people but pur-
course assistant to know volved in the course can sue their studies independently. New
every student’s status divide their tasks among York’s Excelsior College reports that 20
(both time-on-task and themselves and target percent of its students take up to 80 per-
progress through the mod- their efforts to particular cent of staff time, indicating a strong need
ules) in each of the four aspects of course delivery. for human interaction, in contrast to the
courses. By using these tech- The coordinator and the four fac- 80 percent of students who require very
niques, Rio Salado has been able to in- ulty members each receive credit for little interaction.
crease completion rates from 59 percent teaching a single course. Whereas before The Ohio State University (OSU) is re-
to 65 percent while tripling the number the redesign, Southern Mississippi designing its introductory statistics
of students handled by one instructor. needed to staff 16 to 20 sections, the uni- course, which enrolls 3,250 students
The University of Southern Missis- versity now requires the equivalent of each year. In the process, the faculty
sippi’s redesign of its world literature only 5 staffed sections to serve all stu- have come up with a metaphor that cap-
course (http://www.center.rpi.edu/ dents. Thus, by using a coordinated ap- tures a new way to think about online
PewGrant/RD3%20Award/USM.html) proach, Southern Mississippi has more learning environments. OSU has created
moved 16 to 20 face-to-face lecture sec- than tripled the number of students that a “buffet” strategy, which offers students
tions (approximately 60 students each) faculty can handle. an assortment of interchangeable paths
per term into a single 800-student online that match their individual learning
section organized around 4 four-week The Buffet Model styles, abilities, and tastes at each stage of
modules. A course coordinator, responsi- Although all of the models discussed the course. Like the emporium meta-
ble for overall course administration, above have demonstrated that they can phor originated by Virginia Tech, a buf-
manages the team-teaching of four fac- successfully improve the quality of stu- fet suggests a large variety of offerings
ulty members (who each teach one mod- dent learning while reducing the cost of that can be customized to fit the needs of
ule in their area of expertise) and four instruction, each of these models tends to the individual learner.
graduate assistants (who help students be attached to one way of doing things Since students learn in different ways,
with writing and grade their essays). The and treats all students as if they were the even the best “fixed menu” of teaching
faculty members are responsible for con- same. In essence, like the traditional strategies will fail for some students. In
tent, complementary materials, quizzes, classroom model, these course-redesign contrast, OSU’s buffet of learning oppor-
and exams. The faculty team offers mod- models represent a one-size-fits-all ap- tunities includes lectures, individual
ularized course content through a com- proach, albeit a much improved one. Yet discovery laboratories (in-class and Web-
bination of optional-attendance live lec- one of the strongest reasons for using in- based), team/group discovery laborato-
tures and required, Web-delivered, formation technology in teaching and ries, individual and group review (both
media- and resource-enhanced presenta- learning is that it can radically increase live and remote), small-group study ses-
tions. Each module lasts four weeks. Stu- the array of learning possibilities pre- sions, videos, remedial/prerequisite/
dents complete a pre- and post-quiz for sented to each individual student. Thus, procedure training modules, contacts for
each module. Links to additional re- the “right way” to design a high-quality study groups, oral and written presenta-
quired literary and/or critical readings, course depends entirely on the type of tions, active large-group problem-
audio and/or video files, and other students involved. By customizing the solving, homework assignments (GTA
resources devoted to particular authors learning environment for each student, graded or self-graded), and individual and
or themes are provided. Writing assign- institutions are likely to achieve greater group projects. Thus, for a specific objec-
ments are administered by WebCT learning successes. tive, students may choose to hear and dis-
and are graded by graduate assistants; Students need to be treated like indi- cuss a familiar vivid example in lecture,
multiple-choice exams administered by viduals, rather than homogenous groups, view and read about a real example in an
WebCT after each module provide stu- and should be offered many more learn- annotated video presentation, encounter
dents with immediate feedback regarding ing options within each course. Rather an example in a group problem-solving
their understanding of particular themes. than maintaining a fixed view of what all session, or generate an example through a
Consistent content coverage means students want or what all students need, group project. Students may elect to ex-
that all students have the same kinds of institutions need to be flexible and create plore a concept by working in a data-
learning experiences, resulting in signifi- environments that enable greater choice analysis laboratory, in an individual Web-
cant improvements in course coherence for students. Students differ in the based activity, or in a facilitated study

36 EDUCAUSE r e v i e w
September/October 2003

 All five models treat the course not as a “one-off” but rather as a set of
products and services that can be continuously worked on and improved.

session or by explaining the con- using a common syllabus, text- practice. But with its connotations of
cept to others. book, set of assignments, and words like regulate, regiment, and homogenize,
To promote commit- course Web site. Students the word standardize does not precisely
ment to follow-through were placed into cohort capture what is required. What higher
and to enable efficient groups of 60 and, within education needs is greater consistency in
tracking of their progress, these groups, into peer academic practice that builds on accumu-
students enter into an on- learning teams of 6 stu- lated knowledge about improving quality
line contract that captures dents each. The re- and reducing costs.
their choices of learning designed course includes All five models discussed above—sup-
modes at the beginning of six modules, each designed plemental, replacement, emporium, fully
each of four units of study. Stu- by faculty experts. A structured online, and buffet—treat the course not as
dents receive an initial in-class orienta- buffet of learning experiences tied to a “one-off” but rather as a set of products
tion that provides information about the each content module was developed to and services that can be continuously
buffet structure, the course content, the meet the varying needs of students with worked on and improved. Two factors in
learning contract, the purpose of the different learning styles as measured by the design strategies used by each model
learning styles and study skills assess- the Myers-Briggs Type Indicator instru- are key: (1) the collective commitment of
ments, and the various ways that they ment. Options for learning included live all faculty teaching the course, and (2) the
might choose to learn the material. Out of lectures and discussions, taped lectures, capabilities provided by information
class, they complete online learning labs and other hands-on experiences, technology. Would it be possible for a sin-
styles and study skills instruments and textual-based material, practice exams, gle professor conducting an online class
receive a report of their results, as well as commercially produced videos, Web- to develop such creative, comprehensive,
directions on how to use this information based resources, and learning experi- learner-centered designs as exemplified
to build the online course contract. ences related to the arts in the students’ by the redesigns discussed above? Per-
Students are initially given a set of de- home communities. haps, if the individual spent most of his or
fault, software-generated study options to FGCU has discovered two things: (1) her career working on the class. Would it
match their learning styles and study the students did not attend any of the live be possible for institutions to offer a buf-
skills; these options can be changed ac- learning experiences, sticking instead fet of learning opportunities to thou-
cording to students’ preferences. The fin- with the text and online materials in sands of students annually without the
ished contract gives each student a de- WebCT; and (2) they did very well—better aid of information technology? Most cer-
tailed listing of what needs to be than the students who attended lectures tainly not. Information technology en-
accomplished, how this relates to the in the face-to-face courses. The average ables best practices to be captured in the
learning objectives of the unit, and when score on standardized exams in the tradi- form of interactive Web-based materials
each part of the assignment must be com- tional course was 70 percent, versus and sophisticated course-management
pleted—leading up to the unit test three 85 percent in the fully implemented re- software. Faculty can add to, replace, cor-
weeks later. Based on their own experi- design, and the percentage of D and F rect, and improve an ever-growing, ever-
ences in the initial unit and on reading grades went from 45 percent in the tradi- improving body of learning materials.
students’ testimonials from earlier quar- tional to 11 percent in the redesigned Sustaining innovation depends on a com-
ters, students may decide to make course. As a result, FGCU plans to mitment to collaborative development
changes in their contracts for subsequent eliminate some of the live course ele- and continuous quality improvement
units. The course software monitors stu- ments and build on the strengths of the that systematically incorporates feedback
dents’ progress on an individualized online materials. from all involved in the teaching and
basis throughout each unit, suggesting al- learning process. e
ternative learning strategies when Conclusion
needed. Currently in higher education, both on Notes
Among the many advantages of the campus and online, we individualize fac- 1. See the “No Significant Difference Phenomenon”
Web site, (http://teleeducation.nb.ca/
buffet model is that it allows research- ulty practice (that is, we allow individual nosignificantdifference/), which provides se-
driven decisions to be made about indi- faculty members great latitude in course lected entries from the fifth edition of Thomas L.
vidual course elements. Florida Gulf development and delivery) and standard- Russell’s 1999 book The No Significant Difference
Phenomenon, a comprehensive research bibliogra-
Coast University (FCGU) has redesigned ize the student learning experience (that
phy on technology for distance education cover-
its required fine arts course (http://www is, we treat all students in a course as if ing 355 research reports, summaries, and papers.
.center.rpi.edu/PewGrant/RD3%20Award/ their learning needs, interests, and abili- 2. William F. Massy and Robert Zemsky, “Using In-
FGCU.html) using the buffet model. ties were the same). Instead, we need to formation Technology to Enhance Academic Pro-
ductivity,” Educom NLII white paper, 1995,
Twenty-five sections of 30 students each do just the opposite: individualize stu- (http://www.educause.edu/ir/library/html/
were consolidated into a single section, dent learning and standardize faculty nli0004.html).

38 EDUCAUSE r e v i e w
September/October 2003