J O U R N A L F O R N U R S E S I N S T A F F D E V E L O P M E N T • Volume 19, Number 3, 130–142 • © 2003 • Lippincott Williams & Wilkins, Inc.

U se of new nursing knowledge

has improved nursing practice and
patient care outcomes. However,
the busy bedside duties nurses
face constitute barriers to reading
and evaluating research. The
nurse may want to use new
knowledge about pain assessment,
wound care, preoperative fasting,
and depression screening to
improve practice but may lack the
skills and know how to evaluate
CRITICAL ANALYSIS OF RESEARCH
the research critically. Barriers to
analysis of the research include
PAPERS
confusing technical terminology,
statistics, and unclear research
reports. The nurse needs to
understand and examine the steps,
cohesiveness, merits, and
Sharon Valente, PhD, RN
weaknesses of the research to
assess his or her level of
confidence in the findings. This
article describes the criteria for
analysis of the sections of a
research report and illustrates
ways that sound research has
improved patient care.
R esearch in a clinical practice
profession enhances schol-
arly knowledge, improves
practice, and helps clinicians de-
liver care with quality, efficiency,
retention of nurses; research offers
educational perks that enhance
professional growth and develop-
ment. Nurses want to show admin-
istrators they are cutting-edge pro-
and effective outcomes (Dunning, fessionals with a practice based on
1999) Participation in research is the research evidence, but they
often a requirement for perfor- need to upgrade their competence
mance evaluations and promo- in evaluating research. However,
tions. Nurses who want to improve many nurses lack experience with
their care, show their effective research terminology, critique,
practices, and advance in nursing methods, and findings. Reading re-
must read and use research. Re- search requires learning a new lan-
search also provides evidence for guage because scientific terms and
effective practice and demon- concepts appear confusing, and
strates a need for a change in prac- nurses confess that research seems
tice (Collins & Robinson, 1996; dull, boring or “it’s all Greek to
Dunn, Crichton, Roe, Seers, & me”—particularly the statistics.
Williams, 1997). Nursing research This article offers strategies for
has led to significant improve- developing competence in read-
ments in nursing care (see Table ing, understanding, and critically
1). Research opportunities also im- analyzing the research. The focus
K E Y W O R D S
prove marketing, recruitment, and is on identifying nursing problems
ANALYSIS
RESEARCH CRITQUE
RESEARCH UTILIZATION
Sharon Valente, PhD, RN, is Assistant Professor at the University of Southern California, Los Angeles,
California.

130 May/June 2003

 T A B L E 1
Research Findings That Changed Practice
Topic Summary Citation

Parents Visiting Hospitalized Pillitteri studied children’s response Seidl, F. W., & Pillitteri, A. (1967).
Children to having parents visit compared Development of an attitude scale
to a traditional limited visiting on parent participation. Nursing
schedule; more liberal visiting Research, 16(1), 71–73.
policies improved childrens’
recovery.
Pain Evaluation and Management Ferrell studied pain in oncology Ferrell, B. R., & Rivera, L. M.
patients and set standards for (1997). Cancer pain education
improved treatment of pain. for patients. Seminars in On-
cology Nursing, 13(1), 42–48.
Sexuality in Patients With Shapira studied women who Roper, J. M., Shapira, J., &
Alzheimers Disease cared for their husbands who Chang, B. L. (1991). Agitation in
had Alzheimer’s Disease and demented patients. Journal of
identified the hypersexual behavior Gerontological Nursing, 17(3),
of Alzheimer’s patients and 17–21.
management strategies.
Violence in Rape Trauma Burgess studied women who had Burgess, A. W. (1985). Rape
and Serial Murders been raped and defined the rape trauma syndrome: A nursing
trauma syndrome and nursing diagnosis. Occupational Health
care needed. Her research Nursing, 33(8), 405–406.
changed legal standards for a
woman’s defense. Her forensic
research on murders established
the profiling system for identifying
serial killers.
Nurse Practitioners’ Care Delivery In 21 studies comparing primary Horrocks, S., Anderson, E., &
care by nurse practitioners with Salisbury C. (2002). Systematic
MDs, there was no difference in review of whether nurse
healthcare, outcomes, cost, and practitioners working in primary
quality. An NP clinic is more cost care can provide equivalent
effective than other clinics or an care to doctors. British Medical
MD on call. Patients were more Journal, 324(7341), 819–823.
satisfied with NP’s education and
counseling.
Nurse Midwives’ and Residents’ In studies comparing the quality of Diers, D. (1981). Nurse midwifery
Obstetric Care routine obstetrical care from nurse as a system of care. In L. Aiken
midwives and residents, results show (Ed.), Health policy and nursing
major reductions in prematurity, practice (pp. 73–89). New York:
neonatal mortality, percentage of McGraw-Hill.
low birthweight, and increases in
symptom-free babies. Nurse mid-
wives provide a cost savings.

and analyzing the problems, meth- these results by chance). Other ically. Some novices hesitate to
ods, data collection, and results. terms may be unfamiliar (e.g., critique a senior investigator be-
One of the first challenges in quasi-experimental design). Learn- cause they respect the senior’s
understanding research is learning ing the terms and specific word knowledge or wisdom. However,
the language of science (see Table definitions is an important step to- it is the researcher’s cohesive and
2). Researchers use common ward comprehending research. logical argument that requires cri-
words, such as “significance,” but When reading research, nurses tique. The nurse also questions
they apply a very narrow defini- need to be critical of the research whether this study, its proce-
tion (e.g., significance ⫽ a statistic project and expect the author to dures, and its findings would
showing the probability of getting “show “each step clearly and log- apply to clinical practice. For ex-

JOURNAL FOR NURSES IN STAFF DEVELOPMENT 131

 T A B L E 2
Frequently Used Research Terms and Definitions
Research Terms Definitions Example

Hypothesis A statement of anticipated relationship A brief educational presentation about

between two or more variables that advance directives will increase the
may be tested empirically. Directional number of patients who complete
hypotheses specify the direction of the advance directives.
relationship (e.g., as exercise
increases, weight will decrease).
Research Question A statement of the topic under investigation. What are the factors that influence
completion of advance directive
documents by patients?
Method The research steps, procedures, strategies, In a quasiexperimental study, a pre-
and activities used to gather and analyze and postsurvey and retrospective chart
the data in a research study. review will be used to document presence
of advance directives in a control
and experimental patient teaching group.
Research Design The overall blueprint or plan for collecting A quasiexperimental design is effective
and analyzing data; it includes the para- for studying the impact of patient
meters or specifications for increasing the education on patient’s knowledge and
internal and external validity of the study. completion of advance directives.
Significance A statement of the importance and value Advance directive education will
of the study to the profession, knowledge decrease unwanted, distressing, and
development, research, and theory. intrusive treatments, reduce family bur-
den in making choices for a loved one,
and improve patient’s autonomy at
end of life. Avoiding unwanted and
highly technical treatments is a cost
savings for families.
Significance (statistical) A statement of the probability that the results In a comparison of people receiving
could have occurred by chance. a teaching intervention about advance
directives with a control group, the
patient teaching group had an average
of 1 advance directive on file as
compared with controls (x ⫽ 0.24).
Findings were significant at p ⫽ .01.
Variable A characteristic or attribute of a person or Knowledge and completion of advance
object that takes on different values within directives are the variables investigated
the population under study (e.g., pain, in this study.
fatigue, depression, knowledge).
Independent Variable The variable (e.g., treatment or intervention) Patient teaching about advance directives
that is manipulated by the researcher. In a is the intervention or the independent
descriptive study, this may be the variable variable in this study (or the cause).
that comes first.
Dependent Variable The outcome or result of the treatment or Knowledge and completion of advance
intervention, also the criterion or outcome directives are the dependent variables or
variable. outcomes.
Reliability of The extent to which an instrument is Blood pressure or temperature are
Measurement Validity a consistent measure over time. reliable if they are consistent. A pulse
of Measurement The extent to which an instrument oximeter is valid if it accurately
accurately measures what it is measures oxygenation.
supposed to measure (e.g., a scale
accurately measures weight).
Conceptual Framework The researcher provides a rational Education and learning are
organization of how the variables in included in a model of behavior
the study relate to each other. change.

132 May/June 2003

ample, the landmark findings that
a daily aspirin reduced heart at- Categories of a Research
tacks were conducted on men Critique
and not women. A critical evalua-
Participation in research is
tor would note that findings might
often a requirement for Although these sections are stan-
only be applied to the men. In a
dard, authors can organize and
landmark study, Tissot concluded performance evaluations and subtitle them in diverse ways. The
that masturbation leads to insan-
promotions. reader looks for the information in
ity, although he limited his study
these categories when the re-
to psychiatric patients in an asy-
searcher leaves the section unti-
lum (Bullough, 1994). He made
tled. The background may also be
an invalid assumption (e.g., mas-
called introduction or may lack a
turbation causes insanity) and
subtitle. Results may be labeled
used a biased sample. Research pharmacological treatments should
findings. The author may expand
studies are rarely infallible; they have expertise in pain manage-
the detail of some topics more
have limits and advantages. Find- ment and pharmacology. A study
than others. A research article will
ings may not apply to all settings. on nursing care should involve in-
contain more detail on methods
Common problems include the vestigators with nursing expertise.
and procedures and analysis,
lack of an ethnically diverse pop- An article may include a brief bi-
while a clinical journal will proba-
ulation, use of English speakers ography of authors or may pro-
bly explain the research questions
only, and use of only college stu- vide scant detail about qualifica-
and illustrate how the results can
dents or volunteers. The re- tions. The reviewer looks for
be applied in practice. Reading the
searcher’s rationale for the sample evidence of the researcher’s com-
original article or primary source
selection should match the study petence throughout the study and
rather than relying on an abstract
purpose and plan. For example, a for accuracy or inconsistencies in
or another author’s review of liter-
study using a veteran population the report.
ature or summary of existing liter-
that was 98% male would be ap- The title, abstract, and report
ature is essential for an accurate
propriate if they are the clinical should be clear and accurately de-
review.
population. pict the problem under study. A
The purpose of the research cri- poorly worded title makes it diffi-
tique is to evaluate the steps, co- cult for scholars to locate the Background of the
hesiveness, merits, and weak- topic in a computerized index, Study/Introduction
nesses of the study. The critique confuses readers, and the result is
should contain a statement of the the article is indexed in an obtuse This section usually follows a brief
purpose, design, methods, and re- way. For example, a computer- abstract and contains the review of
sults of the research before a dis- ized literature search for articles the literature. The authors typically
cussion of strengths and weak- on suicide among children pro- state the problem clearly, define
nesses. The goal is to develop a duced an amazing array of articles their terms, and describe the ques-
balanced evaluation of the study on “children” who were adults tion under investigation. They il-
and its contribution to scholarly age 20 to age 55 and nothing on lustrate the importance of the re-
knowledge and practice. suicide among those under age search problem or topic to
12. In the articles from the litera- knowledge development, practice,
ture search, suicide occurred dur- or research.
Evaluation Criteria ing adulthood, but the titles may The research problem identifies
have reflected the bereaved par- the topic under investigation that
ents’ perspective that “my child may arise from clinical practice,
The researcher’s qualifications to committed suicide.” The reader the research literature, or theory.
conduct this type of research may who looked for an article on sui- The reader expects a clear depic-
include expertise in the topic, ed- cide among children would be tion of what the study is about,
ucational degrees, credentials, surprised to discover the articles why it was done, and how it was
and previous publications. Con- were about “adult children” be- carried out (Fleming & Hayter,
sultants often have expertise in cause the title suggested other- 1974, 1975). A nurse may test a
such specialized areas as statisti- wise. A brief summary of the sec- theory of pain or complicated grief
cal analysis or cultural issues. Re- tions of a research report is found or apply a nursing theory such as
search teams studying pain and in Table 3. Orem’s Self-Care theory or Roy’s

JOURNAL FOR NURSES IN STAFF DEVELOPMENT 133

 T A B L E 3
Overview of Sections of a Research Report
A determination of the merits and worth of the study reflects several dimensions,
and a research report typically contains these sections.
Background of the Study—explains the problem under investigation within the
context of the current research. The gaps in the published literature provide a
foundation for the study. The discussion introduces and defines the major
variables to be studied, the theory or framework for understanding relationships
among variables, and the question to be studied. If the researcher investigated
pain, the background would introduce pain and its aspects under study and
explain how the proposed study will answer important questions and contribute
to knowledge and practice.
Method—describes the design, sample, data collection procedures, and analysis. If
the study involves tools or instruments, an explanation of their consistency or
reliability and accuracy is included.
Results—capture the findings of the study for each research question or hypothesis.
Discussion—offers an interpretation of the results and compares them with other
research findings and their relationship to practice, theory, policy, and
research.
Implications and Application—examines ways that the findings of the study can be
applied to practice and the implications of the findings for further research.

Adaptation Theory in the research research topic or problem sets the to nursing and patient care. A
study. Theory-based research stage for the design and methods. study may be important or signifi-
makes a substantive contribution The literature review explains a ra- cant because the problem (e.g.,
to knowledge development. A tionale for the choice of the most pain, depression, infection, or sui-
nurse midwife might study pain re- effective design, methods, and in- cide) leads to substantial morbidity
lief and want to compare the gold struments. An outdated or skimpy or mortality. Solid arguments for
standard of injectable lidocaine literature review raises questions significance include cost savings,
with a topical anesthetic. The about whether the researcher quality care delivery, prevention of
problem should be clear, research- knew the current scope of the disease or death, knowledge de-
able, and significant. problem. However, common sense velopment, and improved safety
The review of the literature may prevails. For example, if the author (e.g., prevent medication errors),
be untitled but it presents the most reviews a topic that was last stud- decreased length of stay, and in-
current research and theory. Typi- ied 30 years ago (e.g., anthrax), creased customer satisfaction. Sig-
cally, most literature is recent, usu- we would expect articles from 30 nificance is not established by say-
ally from the last 5 years, and cov- years ago. The reader who is well ing the problem is interesting. For
ers the most important studies. In versed in the topic may recognize instance, a student wanted to
a historical study or a study with that some classic or landmark study which types of nursing posi-
little recent research, older studies studies or pertinent results were tions had the highest salary. She
are necessary. If most of the refer- not cited. Otherwise, the reader stated her rationale for the study
ences are over 10 years old and asks if the problem, variables, and was to expand her personal
the problem (e.g., teen pregnancy) background appeared organized, knowledge. She missed the
is a frequently researched topic, logical, and cohesive. point—the significance should be
the reader wonders why newer in- to the profession or practice. For
formation is absent. Sometimes the Significance example, the nurse who reads a
literature is briefly summarized. report on bicycle helmet safety,
The goal is to explain the gaps in Although it is often untitled, this smoking cessation, or foot fetishes
the literature and show how this section should provide a clear, or- should not have to guess how this
proposed study will expand ganized, and cohesive argument relates to nursing or how the study
knowledge. The definition of the for how the problem is significant can enhance patient care.

134 May/June 2003

Research Questions and the population and the selection
Hypotheses Guide the Study criteria for the people or informa-
When reading research, tion that will be studied. The re-
A research question occurs fre- search design should follow logi-
nurses need to be critical of
quently in descriptive studies or cally from the background and be
when scant research exists and
the research project and clear and appropriate for the prob-
does not clarify the important vari- expect the author to “show “ lem. Each design has strengths and
ables. Sometimes the research weaknesses that should be de-
each step clearly and
question is hidden within the pur- scribed. The researchers should
pose of the study. If the literature
logically. provide a rationale for the selec-
identifies the variables, then the tion of the design. Descriptive
researcher may pose a hypothesis studies are useful if little is known
or a best estimate about how two about a subject or if qualitative ap-
or more variables correlate. For of death. Her study did not apply proaches are the effective way to
example, the concept of compli- or relate to a nursing theory (Saun- examine experiences. For an ex-
cated grief is poorly understood so ders, 1982). If the researcher is perimental design, the reader asks
a research question might be, studying the relationship among if the potential influences on the
“What are the psychological and variables, a theoretical framework results are well controlled. If a
physical characteristics of people showing how the variables relate control group is used, is it the ap-
with complicated grief?” If, how- is useful (e.g., the relationship of propriate control?
ever, two types of face lift (or exercise and metabolism to obe- Some designs use meta-analy-
childbirth method) are both well sity). Alternatively, the researcher sis, a technique to re-examine and
respected but we do not know may describe the relationship analyze data from prior research
which is most effective, the hy- among variables using a concep- studies to determine the results
pothesis might be “Patients who tual framework, a set of abstract over many studies. Other studies
receive face lift A versus B will concepts that broadly explains the use a consensus approach to re-
have greater satisfaction after phenomena of interest and that il- viewing the literature and summa-
surgery, less discomfort and lustrates or maps how variables re- rizing the conclusions. In each
swelling, and faster healing.” Of late. A researcher might apply study, the problem, procedures,
course the type of face lift would models including Peplau’s Inter- and design are evaluated for
be precisely described. personal Relations in Nursing and strengths and weaknesses. The
Orem’s Self-Care Concepts of Prac- reader asks, “Is this the best de-
tice. In a study of how exercise re- sign for the study; does it allow
Theory or Theoretical lates to wound healing or immune enough control and flexibility; and
Framework function, a conceptual framework what are the limits of this design
shows the relationship among and the effectiveness of execution
Some researchers use a theory or variables (e.g., exercise, healing, of the design?” Other questions in-
theoretical framework for their re- immune function). clude, “How will you get your
search and relate all the variables, The background sets the stage sample; what will you use to se-
measures, and hypotheses to this for whether this will be a histori- lect the sample; what are the cri-
framework—often nursing theory cal, naturalistic, or descriptive teria for including someone in the
such as Roy’s theory of adaptation study, a replication of a prior sample?” The issues of control of
in patients recovering from stroke study, or a quantitative study. outside influences (extraneous
or Orem’s Self-Care theory in can- variables), sampling, measure-
cer patients is used. A research ment, analysis, and ethical issues
study does not necessarily need a Method: How the Study require discussion. Some authors
theory, but many studies benefit with multiple publications will cite
from a framework illustrating how
Was Designed and
an earlier publication that details
the variables relate to each other. Carried Out the method or instruments. This
Alternatively, some research stud- can be frustrating and irritating to
ies are atheoretical and gather de- This section describes the blue- the reader who is trying to critique
scriptive data in order to establish print or design for the study and a study.
a theory to explain the problem. the procedures for sample selec- In general, designs tend to be
For example, Saunders studied tion, data collection, and analysis naturalistic or qualitative or con-
widow bereavement after modes (see Table 4). The authors explain trolled (quantitative) and some re-

JOURNAL FOR NURSES IN STAFF DEVELOPMENT 135

 T A B L E 4
Types of Design
Design Characteristics of Study Illustration

Descriptive Research/ To depict the characteristics of individuals, A descriptive study of health

Nonexperimental groups, or phenomena of interest seeking behavior among
college students who have
had unprotected sex
Qualitative Design A naturalistic approach to gathering, An observational study of the use
organizing, and interpreting non- of restraints on inpatients; a
numerical information to understand field study; narratives of women
a person’s or group’s experience. who report their reactions to rape
Quasiexperimental To examine the results of an experimental A study of the impact of music
intervention without meeting the criteria and foot massage on reducing
of a true experiment (e.g., subjects cannot nausea and pain (no control
be randomly assigned or controlled) group was used)
Experimental The investigator aims to test an intervention Randomized clinical trials of
or treatment with rigorous control by two different drug treatments
randomly selecting subjects and controlling the for obesity
independent variable.
Clinical Trials Describing phenomena by manipulating Statistical analysis of randomized
numerical data to determine the extent drug trials
and degree of relationships among the
variables
Case Study A research method that uses a thorough, A case study of a nurse’s
in-depth analysis of an individual, or group. response to the suicide of
a patient
Retrospective To link a current problem with earlier A survey of people with HIV about
influences on the problem (e.g., a study their earlier behaviors (e.g., safe
of medical records of suicides after their sex, number of partners, blood
deaths) transfusion, IV drug use); a
chart review of antecedents of
completed suicide
Meta-analysis A statistical approach to reviewing A meta-analysis of critical
several research studies, combining incident debriefing studies
their results, and analyzing the results
of multiple studies on a topic

searchers use a combination. The Sampling is often an untitled sample. Selecting only students
naturalistic study investigates the section that refers to the relation- who received an A grade to evalu-
topic in its natural setting without ship between the individuals, ate the effectiveness of an instruc-
controls. Anthropologists, social group, or items to be studied and tor’s teaching abilities would likely
scientists, and nurses often use this the larger population they repre- be a biased sample. If the goal of
method of investigating a person’s sent. The authors explain the crite- the study is to generalize the find-
experiences or lifestyle. The quan- ria for the sample and who will be ings from a small group of people
titative study creates a simple and included and how they will be lo- to the larger population, then
very controlled situation and mea- cated. The reader questions quantitative researchers prefer ran-
sures a precise aspect of the situa- whether the sample will allow the dom sampling. Random sampling
tion. A good example is a clinical researcher to answer the research gives everyone in the population
drug trial. More recently, studies question or test the hypotheses. of interest, or target population, an
have used both quantitative and Another question is whether the equal chance of participating. If
qualitative approaches to obtain a sample is representative of the we studied breast cancer in
broader perspective of the prob- larger population of all people women, and wanted a random
lem. Types of research design are with the problem and whether any sample, every woman of diverse
listed in Table 4. bias exists that may influence the age or ethnicity would need a

136 May/June 2003

chance to participate. Often, the wants to know when, where, how,
researcher uses a convenience and how often these data will be
sample of those who volunteer collected and the accuracy and re-
from a healthcare setting. If the re- liability of data collection. For ex-
searcher only studies women over Data collection measures are ample, if researchers survey pa-
age 18 who speak English, the re- tient satisfaction with nursing care,
a “yardstick” that needs to
sults cannot be generalized to non- they should explain the accuracy
English speaking or younger measure consistently. and reliability of the measurement.
women. Although quantitative re- Data collection measures are a
searchers prefer random sampling, “yardstick” that needs to measure
it is not always the most appropri- consistently. If the yardstick were
ate or feasible approach. Alterna- not tested or calibrated, it might
tive sampling strategies exist. For have the properties of a rubber
instance, when one studies les- survey’s specific goals, but it pro- band and measure haphazardly.
bians or a hidden group where the vides a national and large dataset Another issue is the compe-
total number of the population is without the challenges of data col- tence, training, and appropriate-
not known, one might use net- lection. For example, the ness of the data collectors. The re-
work sampling. Network sampling health/weight standards were de- searcher needs to detail their
may involve distributing flyers veloped from insurance company training and qualifications to show
inviting people to participate in a statistics collected from mainly that the data collectors were care-
study and asking those who see white males who had life insur- ful, accurate, knowledgeable, and
the flyer to give it to their associ- ance policies. These gold stan- trustworthy. In one study, I care-
ates. dards were generalized to many fully trained the students in gather-
Next, the reader considers the diverse populations. Although so- ing data from the chart records
sample size of the study. In a ciologists and public health and and had the students rate their per-
quantitative study, generally a health services researchers use cent agreement on 33% of charts.
large sample is preferred. Al- large and previously collected All went well until I verified one
though a survey of 500 or 1,000 datasets, few nurses have done so. student’s data and discovered that
nurses provides a broader view of Replicating a previous study and she was “guessing” at the patient’s
different nurses’ responses than a analysis of existing datasets also income—she recorded that a pa-
study of 12, the qualitative study of offers useful and efficient ap- tient on state disability earned
12 might provide more in-depth proaches for nursing research. $40,000 yearly. Her unreliable data
information. However, some types Consistently throughout an ap- were discarded.
of naturalistic or qualitative re- plied research article the nurse A description of the instruments
search are designed to study small asks, “Will this study gather useful in the section on Method(s),
groups. The question is whether information about a problem that is should include their reliability or
the sample will provide the suffi- important to nursing?” and “Will it consistency, validity, and trustwor-
cient data. Researchers can use a improve nursing practice, research, thiness. Often the measures are
large database, such as the na- or theory?” Applied research is pain or symptom rating scales, at-
tional crime statistics or national practical research and is a scientific titude, teaching, or patient satisfac-
suicide statistics to study a large study that aims to generate knowl- tion measures or knowledge rating
population, a statistically less com- edge that can apply to practice. If scales. The researcher needs to de-
mon problem or national trends in the researcher neglects to provide scribe each instrument, its use in
healthcare data. These large enough information about the sam- this population, and its statistics
datasets help answer questions ple, this is a weakness. for reliability and validity (accu-
about healthcare trends such as, racy to measure what it says it
“Are more C-sections performed in measures). Failure to do so is a
certain parts of the country?” or Data Collection Procedures weakness. The reader is not ex-
“How do patients rate their health pected to “guess” how well the
care from and satisfaction with The procedures for gathering the measure worked or its properties.
nurse practitioners versus physi- data may include surveys, observa- Sometimes the wording of a ques-
cians?” Analyzing an existing set of tion, questionnaires, biophysio- tion or the order of research in-
data is an intriguing opportunity. logic measures (e.g., blood pres- struments makes a big difference.
On one hand, analysis is limited to sure, lab assays), interviews, and This is true in research in different
the original data collected for the standardized tests. The reader language groups. For example,

JOURNAL FOR NURSES IN STAFF DEVELOPMENT 137

one researcher reports translating gathered, recorded, and then inter-
an English language survey of preted.
young women’s knowledge and
use of contraceptives into another
language. The word “condom” Critical evaluation of research Protection of Human Subjects
was translated as “a pocket of
is a skill that requires time to
love,” which was an idiosyncratic Recently, researchers have placed
term that no one understood and develop and refine. more emphasis on documenting
would have invalidated the ques- the risks and benefits, harms, and
tion if not corrected. Procedures safeguards for human subjects. Al-
for translation and reverse transla- most all academic and healthcare
tion typically solve this problem. A institutions require that all research
pilot study testing the instruments on living beings be reviewed for
and measures helps solve some of researcher presents the statistics protection for living subjects, both
these dilemmas. The researcher that support but do not prove the human and animal. The researcher
needs to acknowledge the limits of hypothesis. When the researcher should show the precautions for
the design. sets the limit for the risk of results protecting human subjects and
occurring by chance, this is called meeting ethical research standards.
the level of significance (p ⫽ .05 or Often the researcher notes that the
Analysis p ⫽ .01 are common). This num- study received Institutional Review
ber and the concept of statistical Board (IRB) approval and that the
Although some readers are significance can be confusing. The subjects provided informed con-
tempted to skip statistical analysis simple way to remember its mean- sent. The reader needs to ask if ad-
and go straight to the conclusions, ing is to subtract the number (ig- equate procedures were followed
this section is important. The re- nore the decimal) from 100—the to ensure that the subjects under-
searcher explains the procedures result is 95—or the researcher can stood and consented to the study.
used to analyze the data for each be 95% confident that the results Researchers should provide a ra-
research question or hypothesis. are a true finding and not an error. tionale if women and minorities
These analytical procedures may This number identifies the risk the were excluded and should explain
be descriptive or inferential. De- researcher is willing to take that how they protected the rights and
scriptive data (e.g., frequencies, the null hypothesis (e.g., no differ- safety of any vulnerable subjects
percentiles, totals, averages, range ence between groups) will be re- (e.g., children, cognitively im-
of scores and their distribution) jected when it actually was true. If paired persons, prisoners, elderly
help depict the characteristics of the researcher wants to know if persons, or stigmatized group).
the sample and count narrative re- knowledge improves after a lec- The researcher needs to make sure
sponses. These numbers may be ture, 95% confidence is acceptable. that a subject can stop participat-
displayed in a table (e.g., educa- If, however, the topic under inves- ing in the study at any time, and
tion, age, ethnicity, income, reli- tigation is life threatening such as that explanations of the study are
gion, and occupation of the sam- the evaluation of suicide risk, the provided in lay language (e.g., we
ple). Descriptive statistics do not researcher will want to be 99% will take 3 teaspoons of blood—
inform the reader about any cause confident or have a higher signifi- not 18 ml of blood).
or effect relationships, but they il- cance (p ⫽ .01 level).
lustrate qualities or characteristics In experiments, the researcher
of a person or phenomenon. Table provides a formula for statistical Results
5 provides a brief summary of analysis of the data for each hy-
common statistical tests. pothesis. Often a statistical consul-
Inferential statistics allow the re- tant helps define the analytic tests. The researcher repeats the hy-
searcher to draw cause and effect To compare the average scores of potheses or research questions be-
conclusions and to relate the sam- two groups of subjects to deter- fore stating the data analysis and
ple findings to the target popula- mine if a difference existed on test findings. Commonly, the results
tion. These statistics reflect the scores ranging from 0–100, the re- will include a description of the de-
probability that the results in the searcher uses a t-test. In a qualita- mographics of the sample. Next,
study occurred by chance. As the tive research study, the researcher the statistical analysis will be re-
risk of chance diminishes, confi- illustrates how the interviews or ported. However, in a qualitative
dence in the results increases. The narratives or observations will be study, the results may not be statis-

138 May/June 2003

 T A B L E 5
Common Statistical Tests
Categories Illustrations Description and Limits

Univariate Descriptive Statistics Measures of central tendency Describe the sample or characteristics
(averages) of the sample; frequencies are often
Frequencies used in qualitative and quantitative
Percentiles studies to count responses
Range

Inferential Statistics Test hypotheses May allow generalization of findings

from a random sample to a population
Chi Square Test of Tests whether two variables being Can use if there is one entry of data
Independence examined are independent or related per subject in the sample—not if the
subjects have repeated measures
T-Tests Measure the difference between two Must meet criteria for t-test that
group means; describe how different includes the type of data you have
the two groups are on a specific and a normal distribution of
characteristic; parametric test used to data
analyze interval data for independent
groups
Pearson’s Correlation Common test used to measure the 0 shows no relationship,
extent of linear relationship .1 – .3 shows a weak relationship,
between two variables and .3 – .5 a moderate and above
.5 a strong relationship; a positive
number shows the increase of one
variable when the other goes up
Analysis of Variance (ANOVA) Tests for differences among three or (e.g., determines if outcomes of three
more means interventions were different)

Probabilities

Level of Significance Probability measure; identifies the p ⫽ .05

probability of predicting an event
without error and the risk of error;
risk of making a Type I error (rejecting
the null hypothesis when it is true)
Degrees of Freedom Freedom of the score’s value to vary
given the other existing scores df ⫽ 10
Power Analysis and Effect Size Statistical analysis can indicate how A sample of 200 will be needed to
many subjects are needed to test a test the hypotheses at 80% level.
hypothesis at a given level of power
(e.g., how certain you want to be).
All statistical tests specify the rules that must be met before the test is valid. For example, the type of distribution and the level
of measurement.

tical. For example, in a study of in- ness of patient care (Valente & This section should omit any dis-
terventions to reduce fatigue, the Saunders, 2000). Knowledge and cussion or interpretation of the
findings might include themes or attitudes were weak predictors and findings.
narratives describing the signs, more research to explain effective
symptoms, and patients’ reports of care was needed. Hence, negative Discussion
their fatigue and what reduced fa- or weak findings can be important
tigue. In a study of oncology (e.g., If the cause is not knowledge
nurses, researchers expected that and attitudes, then education is not The authors compare and contrast
knowledge and attitudes toward the solution.) Tables, graphs, and their results with other studies and
suicide would predict the effective- pictures help display the results. interpret the findings. They may

JOURNAL FOR NURSES IN STAFF DEVELOPMENT 139

 T A B L E 6
Guidelines for Critique
Topic Sample Criteria for Critique Written Critique (Sample)

Problem and Purpose • Is the problem clear, precise, and The authors clearly defined an
well defined? important and underexamined topic
for research.
• Does it match the purpose? The problem, low use of advance
directives, is a costly patient care
problem that places undue burden on
patients, families, and nursing staff.
• Is it current and important to nursing? Understanding use of advance
• Is the problem appropriate and feasible directives is necessary to improve
for research? clinical practice and research, or
• What is the topic under investigation? authors fail to explain the
Significance • Is a good argument made for the significance.
importance of the problem to clinical
practice, research, theory, or knowledge
and policy development?
Literature Review • Citations include relevant and current The literature is over 10 years old
research. and outdated; few references are
from healthcare journals. Many
citations are country or state reports
or Internet sites that may not be
available over time.
• Current knowledge and gaps in
knowledge of the problem are explained.
• Key literature is cited accurately.
• Citations are complete.
Human Subjects • Are the rights of human subjects The authors clearly described
protected? Has the study been approved sound procedures for protection of
by an appropriate IRB? Can the subjects human subjects and informed
give consent and understand the study? consent.
• Are minorities or women excluded Women and minorities are included.
without a solid rationale?
Method • Does the research design match the Research design, problem, and
problem and appear appropriate? methods are sound and well
Would another design have been better? matched. Rationale explains
What are the limits of the design? selection of these approaches.
Interventions • Interventions and controls are described This is a descriptive study; no
in detail. intervention.
Sample • Target population is clear, described, The researchers used a convenience
and appropriate. sample of psychiatric patients at an
• Eligibility or exclusion criteria, recruitment, urban clinic. Rationale for lack of a
and sampling procedures are clear. random sample from diverse settings
• Sample size is appropriate. is absent. Rationale for using English
• Did any of the sample drop out and speakers is absent.
why?
• Were there any biases?
• Was a convenience or random sample
used?
Instruments • The reliability and validity of instruments The Beck Depression Inventory is a
are cited; the rationale for these well-known instrument. Rationale and
instruments is clear and convincing. psychometrics for use in primary care
• Limits to the instruments are clear is absent. Argument is lacking for why
(e.g., will the instruments gather this is the best instrument.

(Continues)

140 May/June 2003

 T A B L E 6
Guidelines for Critique (Continued)
Topic Sample Criteria for Critique Written Critique (Sample)

the data needed; will there be

problems with poor recall, lying
about sensitive information?)
Data Collection • Data collection procedures are clear. Data collection procedures are
• How many times were data collected? unclear. Description is lacking for
• Did the researcher control external frequency of data collection, controls,
influences? and training of data collectors.
• Will the data answer the question?
• Data collectors are trained, reliable,
and accurate
• Circumstances of data collection
were conducive to data collection
(e.g., confidentiality).
• Did the researchers minimize any risks,
stress, distress, or burden to subjects?
Data Analysis • Procedures match the research question, Descriptive statistics describe the
design, and data collection. sample. t tests match hypotheses.
• Statistical tests match the data and However, how error will be managed
hypothesis. with multiple t tests is unclear.
Results • Results are clear, complete, and Results are complete and match
identified; they answer the research research questions. Significance is
questions. stated.
Discussion • Results are compared and contrasted Discussion is absent.
with other research findings.
Conclusions/Implications • Conclusions accurately reflect the data. Conclusions accurately reflect the
• Implications for practice, research, and data. Implications for practice,
theory are discussed. research, and knowledge are absent.
• No unwarranted conclusions or
interpretations are made

confirm other researchers’ findings clude that this also is true for
or contradict them and search for Conclusions/Implications women (who were not studied).
an explanation for the contradic- The implications need to describe
tions. If these study findings con- The researcher summarizes the how the research findings can be
tradict earlier results, the re- major findings and conclusions applied to nursing practice, can
searchers discuss this. The and their application to practice, suggest future research studies,
researcher can also examine trends research, theory, and knowledge and can improve theory and
in the data that were promising but development. Implications for fur- knowledge.
did not reach significance and ther research are mentioned. Cau-
speculate why the trends were not tion in generalizing the findings Summary
significant. The limits of the study beyond the data is emphasized.
are outlined. The researcher may The reader asks whether the con-
also discuss problems encountered clusions represent the data accu- Critical evaluation of research is a
during the research such as the dif- rately. Often even the most senior skill that requires time to develop
ficulty obtaining an ethnically di- researchers can draw a conclusion and refine. In addition, some re-
verse sample despite intensive ef- that is not based on the data even search reports simply provide less
forts to do so. If a natural disaster though the conclusion may appear than ideal information for the cri-
occurs during a study on depres- plausible. For example, if the re- tique. If the researcher does not
sion, the authors need to discuss searcher studies men and con- explain the sample criteria, ratio-
the impact of the disaster on their cludes that aspirin reduces risk of nale for sample selection or some
findings. stroke, the researcher cannot con- other aspect, the reader cites this

JOURNAL FOR NURSES IN STAFF DEVELOPMENT 141

as a flaw. It may not be possible to Appreciation is extended to Sandi oiku— Japanese Journal of Nurses Ed-
determine if this is just a defect in Strome and Ahnnya Slaughter for ucation, 16(5), 260–265.
the written report or a flaw in the review and editing suggestions. Fleming, J. W., & Hayter, J. (1974).
research design or implementa- Reading research reports critically.
tion. Sometimes, the publisher’s Nursing Outlook, 22, 172–175.
REFERENCES
guidelines and space limit what is Horrocks, S., Anderson, E., & Salisbury
put in print. Some aspects of the Bullough, V. L. (1994). Science in the C. (2002). Systematic review of whether
critique rely on the reader’s judg- bedroom. New York: Basic Books. nurse practitioners working in primary
ment, experience, and common Burgess, A.W. (1985). Rape trauma care can provide equivalent care to
sense. A critique should contain a syndrome: A nursing diagnosis. Occu- doctors. British Medical Journal,
systematic assessment, be unbi- pational Health Nursing, 33(8), 324(7341), 819–823.
ased, and identify the strengths 405–406.
Roper, J. M., Shapira, J., & Chang, B. L.
and weaknesses of the study (see Collins, E. G., & Robbinson, D. (1996). (1991). Agitation in demented patients.
Table 6). The critique can identify Bridging the research-practice gap. Journal of Gerontological Nursing,
whether the research report con- Nursing Standard, 10, 25. 17(3), 17–21.
tained sufficient data for critique, Diers, D. (1981). Nurse midwifery as a Saunders, J. M. (1982). Uncoupled
was well written, well organized, system of care. In L. Aiken (Ed.), identity: A process of bereavement res-
clear, concise, and lacked jargon. Health policy and nursing practice olution. Western Journal of Nursing
Current standards also require that (pp. 73–89). New York: McGraw-Hill. Research, 3(4), 319–332.
authors avoid sexist language, pro- Dunning, M. (1999). Experience, evi-
Seidl, F. W., & Pillitteri, A. (1967). De-
vide citations for strong state- dence, and everyday practice. London,
velopment of an attitude scale on par-
ments, offer complete and accu- Kings Fund.
ent participation. Nursing Research,
rate statistics, and write a concise, Dunn, V., Crichton, M., Roe, B., Seers, 16(1), 71–73.
logical, and orderly report. K., & Williams, K. (1997). Using re-
search for practice. Journal of Ad- Valente, S. M., & Saunders, J. M.
vanced Nursing, 26, 1203–1210. (2000). Oncology nurses’ difficulties
ACKNOWLEDGMENTS with suicidal patients. Medicine and
Ferrell, B. R., & Rivera, L. M. (1997). Law, 19(4) 793–814.
This material is the result of work Cancer pain education for patients.
supported with the resources and Seminars in Oncology Nursing, 13(1),
use of the facilities at the Depart- 42–48.
Address for reprints: Sharon Valente, 346 N.
ment of Veterans Affairs, Greater Fleming, J. W., & Hayter, J. (1975). Bowling Green Way, Los Angeles, CA 90049
Los Angeles Healthcare System. Reading research critically. Kango Ky- (e-mail: Sharon.valente@med.va.gov).

142 May/June 2003