A Practical Guide To The Comparative Case Study Method In Political Psychology
1. A Practical Guide to the Comparative Case Study
Method in Political Psychology
Juliet Kaarbo
University of Kansas
Ryan K. Beasley
Baker University
The case study, as a method of inquiry, is particularly suited to the field of political
psychology. Yet there is little training in political science, and even less in psychology, on
how to do case study research. Furthermore, misconceptions about case studies contribute
to the methodological barrier that exists within and between the two parent disciplines. This
paper reviews the various definitions and uses of case studies and integrates a number of
recent insights and advances into a practical guide for conducting case study research. To
this end, the paper discusses various stereotypes of the case study and offers specific steps
aimed at addressing these criticisms.
KEY WORDS: case study, comparative method, research design.
The field of political psychology has become increasingly populated with
individuals capable of speaking both the language of political science and the
language of psychology. This is valuable for building theory and understanding the
nature of the empirical domains of interest. Methodologically, however, there
remain some persistent obstacles between the two parent disciplines that, in our
view, unnecessarily hinder empirical cross-fertilization and advancement. Both
social psychologists and political scientists, for example, study the structure and
content of mass belief systems, but tend to use different methods (experiments and
surveys) and different statistical techniques (analyses of variance and regressions).
Margaret Hermann (1989) has indicated that methodological differences between
political science and psychology threaten to make the promise of a field of political
psychologymerelyafantasy.Ifwearegoingtocontinuetodevelopaninterdisciplinary
Political Psychology, Vol. 20, No. 2, 1999
369
0162-895X Š 1999 International Society of Political Psychology
Published by Blackwell Publishers, 350 Main Street, Malden, MA 02148, USA, and 108 Cowley Road, Oxford, OX4 1JF, UK.
2. language, methodological dialogue must replace the current tendency toward
simultaneous monologues.
Nowhere is the methodological gulf in understanding wider than when politi-
cal psychologists contemplate case study research (Tetlock, 1983). Indeed, a
review of the articles published in Political Psychology over the past decade reveals
that the case study is one of the most popular methods, but is most often used by
political scientistsâespecially those who study American leadership, international
relations, and comparative politics.1 Psychologists who publish in this journal tend
to use survey, experimental, or content-analytic techniques.2 Psychologists only
occasionally use the case study method.3
Practitioners of case study research, whether political scientists or psy-
chologists, lack a common language for their investigation. Alexander Georgeâs
(George, 1979; George & Bennett, in press; George & McKeown, 1985; George
& Smoke, 1974) method of âstructured-focusedâ comparison is perhaps the most
used resource for those engaging in this methodology. His articulation of the
procedures involved in the case study method have provided a sturdy methodologi-
cal foundation for hosts of empirical investigations. Yet many are less than
forthright in articulating the rationale of their chosen methodology. Indeed, Ragin
(1987) suggested that âmost investigators who use case-oriented strategies . . . are
not self-consciously methodological; that is, they do not regard the case-oriented
strategies they use as formal methodologiesâ (p. 34).
Although there has been a resurgence of attention to the case study method
within a variety of disciplines (see, e.g., Orum, Feagin, & Sjoberg, 1991), we argue
that there has been insufficient attention to and internalization of these develop-
ments within political psychologistsâ parent disciplines. This is true despite the
degree to which many political scientists embrace this technique and the tendency
for many psychologists toeschewit. Our purpose here is tosummarize and integrate
a number of these recent insights and advances into a practical guide for conducting
case study research, thereby providing greater direction and a common language
for those engaging in this type of research and highlighting the merits and value of
this technique for those more inclined to avoid it. Toward this end, we draw on
1 Recent examples of case study research in these subfields include Fishelâs (1992) case study of a
California legislatorâs development over time, Gilbertâs (1995) case study of President Johnson for
the political effects of illness, Steinbergâs (1991) case study of the Cuban Missile Crisis for the role
of shame and humiliation in conflict deterrence, Farnhamâs (1992) case study of Roosevelt during the
Munich Crisis as an example of prospect theory, and Groth and Brittonâs (1993) case study of the
Soviet Union for the psychopolitical foundations of a garrison state.
2 Recent examples of research with these techniques include Fife-Schaw and Breakwellâs (1990) survey
of teenagers to predict the intention not to vote, Milburn and McGrailâs (1992) experiment on the
effects of cognitive complexity of news presentation, and Suedfeldâs (1992) content analysis of
newspaper editorials during times of positive and negative relations between pairs of countries.
3 Exceptions include Ravenâs (1990) case study of the MacArthur-Truman relationship as an example
of interpersonal influence and Tiedensâ (1997) comparative case study oftwoPolish ghettos for violent
resistance during the Second World War.
370 Kaarbo and Beasley
3. numerous writings on the case study and comparative case study methods. We are
not intending to provide an array of novel reasons to use the case study method.
Rather, we are attempting to integrate observations regarding this method into a
coherent program for its use in a research effort, with particular attention to the
field of political psychology. In this sense, we are not comparing the case study
method to other methodologies, but instead are attempting to address various
perceived shortcomings and stereotypically founded criticisms within a practical
guide for actually doing case study research.
In this effort, we explicitly avoid any direct discussion of numerous and
engaging epistemological issues that are part of the current debate in political
science on the value of interpretation and the role of positivism in the social
sciences.4 The case study method is particularly prone to evoke discussions of the
nature of theories, the possibility of general laws governing human behavior, and
other fundamental issues of investigation. Although important, these issues are not
central to our discussion. Instead, our effort concerns the basic decisions and
choices that must be made in the practical application of the case study method.
In both political science and psychology, there exists a stereotype of the case
study that includes an idiosyncratic, historically specific, and atheoretical charac-
ter; a lack of control; an inability to generate a sufficient number of data points to
test theory; and a highly suspect âinterpretiveâ character that allows the analyst to
draw favorable conclusions about hypotheses and to find support where no such
support exists. Certainly, the ways in which case studies have been performed and
used have frequently been rather divergent from the standards of systematic inquiry
currently widely accepted in both political science and psychology.
The reasons that the case study has fallen between the stools of methodological
rigor and empirical value are numerous. Among political scientists, the compara-
tive case study is often equated with cross-national comparison. The writings on
case study research tend to be associated with the subfield of comparative politics
and are not widely read in other subfields. Further, political scientists have paid
insufficient attention to recent discussions of the comparative method and case
study research found in public administration and sociology. Psychologists, on the
other hand, tend to associate case study research with the clinical âcaseâ or patient
analysis and are generally unaware of the political science scholarship on the use
of case studies. Overall, we believe that political psychologists could be made more
aware of the different meanings of case study, the different types or purposes of
case study research, and the steps involved in case study research design, and that
they can make good use of the method as part of a systematic comparative
methodology without necessarily succumbing to its traditional pitfalls.
4 See, for example, the âReview Symposium: The Qualitative-Quantitative Disputation: Gary King,
Robert O. Keohane, and Sidney Verbaâs Designing Social Inquiry: Scientific Inference in Qualitative
Researchâ in the American Political Science Review (1995); George and Bennett (in press).
Comparative Case Study 371
4. Defining âCase Studyâ
The terminology pertaining to âcase studiesâ is frequently confusing. Ragin
(1987) noted that âthe term âcaseâ and the various terms linked to the idea of case
analysis are not well defined in social science, despite their widespread usage and
centrality to social scientific discourseâ (p. 1). For example, a âcaseâ may be a
single instance or data point, such as a subject in an experiment, a survey respon-
dent, or the non-occurrence of war between belligerents (King, Keohane, & Verba,
1994). Alternatively, it may be a uniquely bounded phenomenon in a historical or
geographical sense, such as the case of Munich, the Soviet invasion of Afghanistan,
or the Watts riots. Indeed, the term âcaseâ can have a variety of meanings, each
with important implications regarding empirical investigation.
When the term âcaseâ is joined by other terms such as âstudy,â âanalysis,â or
âmethod,â the picture does not become any clearer. For example, Orum et al. (1991)
defined a case study as âan in-depth, multifaceted investigation, using qualitative
research methods, of a single social phenomenonâ (p. 2). Lijphart (1971, 1975) saw
the case study as a single case that is closely associated with the comparative
method, as contrasted with experimental and statistical methods. George and
McKeown (1985) indicated that a case study focuses on within-case analysis to
evaluate claims about causal process. Finally, Yin (1994) defined a case study as
âan empirical inquiry that: investigates a contemporary phenomenon within its
real-life context, especially when the boundaries between phenomenon and context
are not clearly evident; and . . . [that] relies on multiple sources of evidenceâ
(p. 13). Clearly there are different definitions of cases and case studies, and this
has made systematic analyses of the value and purposes of this technique difficult.
We recognize that coming to some consensual understanding of the terms
âcase,â âcase study,â or, more generally, the âcase methodâ is extremely difficult
(see Ragin & Becker, 1992), yet without some understanding of what these terms
are meant to convey, it is even more difficult to offer any practical guidance in
carrying out an investigation. For our current purposes, we offer basic definitions
of several terms. We consider a case to be an instance, or a data point, and in this
sense we do not draw any distinctions about how a âcaseâ is further defined or
arrived at in an investigation. Cases can be experimentally derived measurements,
survey responses, or classifications of historical events (such as âwar/no warâ). We
define a case study to be a method of obtaining a âcaseâ or a number of âcasesâ
through an empirical examination of a real-world phenomenon within its naturally
occurring context, without directly manipulating either the phenomenon or the
context. The comparative case study is the systematic comparison of two or more
data points (âcasesâ) obtained through use of the case study method.
This definition has a number of implications. First, we do not assume a
particular purpose of the investigation, such as developing causal inferences or
detailing historical occurrences. A case study can be used for a variety of purposes,
several of which we articulate below. Second, we do not assume that multiple
372 Kaarbo and Beasley
5. sources of evidence must be used (as does Yinâs definition above), although we
suspect that this will frequently be necessary. The case study, in other words, often
uses a number of techniques for gathering informationâfrom interviews to surveys
to content analysisâbut it is not necessary to use multiple sources or types of
evidence in order to perform a case study. Third, we offer no particular distinction
between qualitative and quantitative aspects of a case study, nor do we draw a
particular distinction between narrative and analytic aspects (see Abbot, 1992;
King et al., 1994). Indeed, we feel that case studies can be very qualitative and
narrative in form, and they can be very quantitative and analytic in form. Fourth,
the issues of context and manipulation are central to our definition. The phenome-
non is studied as it occurs within its context, and neither the phenomenon nor the
context are directly manipulated by the researcher. Finally, our definition allows
for comparison of several case studies and is not limited to single case studies.5
Types of Case Studies
Our definition of the case study does not imply a single purpose for an
investigation. Indeed, there are many uses of case studies that have been outlined
by a variety of scholars (e.g., Eckstein, 1975; Lijphart, 1971; Van Evera, 1996;
Yin, 1994). In our effort to offer a practical guide for the case study method, we
condense these previous discussions into a single, accessible form. In doing so, we
hope to highlight the potential value of case studies for a variety of purposes as
well as lay the foundation for our arguments regarding the need for greater
systematicity in comparative case studies when theory development or testing is
the goal. We present the different uses or types of case studies as though the
investigator will be using a single case study, although this typology in no way
limits analysts to investigate only one case at a time.
Using Cases for Description
The first type of case study is what Lijphart (1971) referred to as âatheoreticalâ
and what Eckstein (1975) called âconfigurative-idiographic.â Here the analyst is
principally interested in the case itself, and is typically attempting to gain a Gestalt
or holistic picture of the event or phenomenon. Generally, with this use of case
studies, the analyst is more interested in the case than in some theory or hypothesis.
Certainly description requires some preconceived ideas or points of reference that
might be considered the foundation of theory, but the principal purpose of the case
5 Several scholars within the social sciences have written extensively about the importance and purpose
of the comparative method (see, for example, Collier, 1993; George, 1979; Lijphart, 1971, 1975;
Ragin, 1987). The comparative method should, in our view, be distinguished and disentangled from
the issues surrounding case studies and the case study method. Comparison is the very foundation of
scientific methodology. In other words, comparison need not be distinguished from other methods,
but can be considered to be a foundation of methodology itself.
Comparative Case Study 373
6. study does not directly involve that set of preconceived ideas. Rather, the case is
meant to be an end in and of itself. With this use of the case study we often observe
analysts using intuitive and descriptive accounts that attempt to lead to some
general âunderstandingâ of the case.6 The use of the term idiographic (as contrasted
with nomothetic) quite directly indicates that the case will not be generalizable or
that the case study will not seek to establish general rules of behavior.
Using Theory to Explore Cases
In contrast to the âconfigurative-idiographicâ or âatheoreticalâ case study,
a researcher may choose to use some theory or set of hypotheses to direct his
examination of a particular case. This use of the case study is âdisciplined-
configurativeâ (Eckstein, 1975) or âinterpretiveâ (Lijphart, 1971). Here, the focus
is still on the case, but the analyst explicitly uses some theoretical foundation in
order to examine or interpret the case. Although there may be some âfeedbackâ
from the case to the theoryâsuch as whether the theory is inappropriate for
understanding the case, or whether no theory currently exists that effectively
illuminates the caseâthe focus is still mainly on the case. After such an endeavor,
we should know the case in a âtheoretically familiarâ form. Here, however, the
purpose of the analyst may be to provide a mechanism whereby changes can be
effected on a particular organization, institution, or actor that was the subject of the
âinterpretiveâ case study. In other words, the interpretation of the case may be
designed to effect changes along the lines of the given theory. The interest remains
on the case (or the actor or actorsof the given case), but the theoreticalinterpretation
of the case may provide some prescriptive leverage.7
Using Cases to Develop Theory
The third type or use of case studies is what has been termed âhypothesis-
generatingâ (Lijphart, 1971) or âheuristicâ (Eckstein, 1975) case studies. Here the
analyst specifically investigates a case in an effort to develop testable hypotheses.
In this sense, the case study is being used to build theory; the analyst examines a
6 Some might argue that this type of case study should be further divided into âhistoricalâ and
âinterpretive,â with the former being an attempt to present events or circumstances in an effort to
reveal the case, and the latter being an attempt to gain an understanding of the case from the perspective
of the actors involved. We, however, have not opted to divide this particular use of the case study
along these lines, as our intention is to provide a more general set of uses of the case study. Further,
following Ragin (1987), we consider that interpretive approaches to case studies are not in competition
with other types of approaches.
7 Freudian analysis applied to a patient, for example, interprets the patientâs past behaviors and activities
according to a more or less defined theory, often for the purpose of changing the patient according to
that interpretation. Decision analysts may evaluate the past behavior of an organization according to
theories about group performance specifically to offer modifications in the organizationâs composition
or behavior according to that theoryâs prescriptive implications for the case.
374 Kaarbo and Beasley
7. specific set of concepts in order to develop generalizable theory from particular
instances. The selection of cases is based on the nature of the theory that is being
constructed, rather than on a particular case that is otherwise of some interest to
the researcher. The focus, in other words, shifts from being explicitly on the case
to being explicitly on the theory. In these types of case studies, the analyst is
typically looking at fewer variables than would be true for the configurative-
idiographic case study, which uses a more Gestalt approach to the investigation,
and specifically attempts to develop a broad and informed picture of the case. This
is very similar to the sort of âintrospective examinationâ that psychologists often
engage in as they are constructing a hypothesis regarding individual behavior or
cognition. They might, for example, ask themselves âhow do I, as an expert, think
about thingsâ when they are attempting to develop a theory of expert cognition.8
Using Cases to Explore and Refine Theory
Another use or type of case study increasingly focuses the analyst on a specific
theory. This has been termed a âplausibility probeâ by Eckstein (1975), who
suggested that plausibility âmeans something more than a belief in potential
validity plain and simple . . . [and] something less than actual validity, for which
rigorous testing is requiredâ (p. 108). Here the analyst is presumably interested in
performing some labor- or cost-intensive study, and thus is benefited by examining
certain aspects of the theory at stake within the confines of an empirical instance.
It is an inductive feedback device within the context of the particular empirical
domain to which the analystâs theory speaks. Further, this use of the case study is
particularly important with novel hypotheses or theories, such that an existing body
of evidence cannot be referenced in speculating about the plausibility of a suspected
relationship. This assists the analyst in determining whether an empirical instance
of a particular phenomenon can be reasonably found, whether the operationaliza-
tions of key variables are in concert with the data available, and/or whether the
posited relationship or relationships between and among variables are consistent
with expectations.9 In this latter form, the plausibility probe helps to avoid a costly
exploration of an implausible explanation.
8 Using a case study as a âdeviant caseâ (Lijphart, 1971) offers the potential for expanding the nature
or scope of an existing theory. Here the analyst specifically chooses a case that deviates from the
pattern predicted by theory, and then attempts to examine the case in an effort to uncover additional
variables that may be lacking in the original theory. In this sense, the deviant case differs from a
heuristic case study in that a theory already exists that is being examined. A deviant case, however, is
similar to a heuristic case in that both are attempting to establish new hypotheses or propositions. In
other words, the deviant case can elaborate a theory by suggesting new hypotheses or additional
conditions under which a certain relationship may or may not hold. In this sense, the deviant case and
the heuristic case are similar.
9 Goldmann (1988), in an appendix entitled âA Note on the Utility of Weak Theory and Weak Tests,â
suggested that using the case study in a fashion similar to the plausibility probe is of value for âtestingâ
weak theory, in that it can âlead to an improved understanding of the problems and possibilities of
operationalizationâ (p. 227).
Comparative Case Study 375
8. This is very similar to many aspects of the pilot study in experimental methods.
It is rare that an experiment is constructed in the office, run in the laboratory, and
published in a journal in swift order. Experimental psychology is replete with
efforts to probe the plausibility of an expected relationship, and the experimental
method is an iterative cycle that effectively incorporates numerous steps designed
to refine theories and hypotheses before testing them.
Using Cases as Tests of Theory
Case studies can also be used as single tests of an established theory, either to
confirm or infirm (Lijphart, 1971) or to investigate a crucial case (Eckstein, 1975).
The distinction between confirming and infirming rests principally in the nature of
the findings, that is, whether they support the predictions of an existing theory
(confirming) or whether they call that theory into question (infirming). In either
event, the process involves specifying in exacting fashion the variables to be
measured and the hypotheses to be tested. Further, a case must be selected such
that it fits the parameters of the theory that is being investigated. Once a population
is specified (in more or less precise form), then a sampling procedure that maxi-
mizes random selection will tend to offer the most compelling interpretation to the
given results of the study.10 Once performed, the case study will stand as a single
âtestâ of a given theory, and the weight afforded that test in terms of its commentary
on that theory will be a function of the way in which the case was chosen, the theory
specified, the variables operationalized and measured, and the conclusions drawn.
Some cases, it can be argued, offer the potential to make a particularly compelling
statement about the theory involved. This has been termed the âcrucial caseâ study
(Eckstein, 1975) and purportedly offers a critical test of a theory, in contrast to the
âuncriticalâorâroutineâtestofferedbyothercaseexaminationsofagiventheory.Here,
the analyst intentionally selects a case where alternative explanations of a negative
finding(afindingotherthanthatpredictedbythetheory)arenotplausible.Thisrequires
both that the theory can be completely specified in terms of empirical expectations,
and that a case exists such that mitigating factors do not exist. As Eckstein himself
acknowledged, this is rare within the social sciences, although how crucial a case is
can be considered along a continuum that allows more or less boisterous claims on the
part of the researcher regarding the support or lack of support afforded by the test.
Steps to Comparative Case Study Research
Much of the writing in political science on the case study method has focused
on justifying the case study or on the different purposes of case studies. There has
10 For a discussion on the important and deceptively difficult question of populations and case studies,
see Ragin and Becker (1992). See also Yin (1994) for a discussion of statistical generalization
(generalizing to a population) versus analytical generalization (generalizing to a theory).
376 Kaarbo and Beasley
9. been less attention paid to how to do case study research. Even in research methods
textbooks, chapters devoted to the case study give the student few guidelines for
how to construct a research design for the case study method (e.g., Baker, 1988;
Isaak, 1985; Johnson & Joslyn, 1991). Our purpose in this section is to outline for
researchers who choose the case study method the important steps for case study
research. What issues are likely to arise in the design of case study research? What
decisions must the researcher face? What types of evidence are compatible with
case study research? These are the questions addressed in the following guidelines.
In doing so, we want to integrate the advice from a number of fields and offer a
practical guide for conducting case study research.
Although most of these steps to case study research would apply to many of
the above uses or types of case studies and to single as well as multiple case studies,
our treatment focuses on comparative case studies with the purpose of examining
hypotheses as critically and systematically as possible. In other words, although
there is an important role for case studies in interpreting a single historical case and
in serving as a heuristic for theory construction, we discuss the steps researchers
will take when they use the case study method to compare two or more cases in
order to draw implications about an existing theory or single hypothesis. We focus
on this type of case study because it is the furthest afield from the âstereotypeâ of
case studies as idiosyncratic, historically specific, and atheoretical. We do not
intend to suggest that the other uses of the case study are not of value, or that many
of these steps could not be applied to improving the other types of case studies.
Rather, we find it least confusing to present these steps with a particular use of the
case study in mind. However, many of the issues we discuss are not necessarily
limited to this purpose of the case study.
To date, the best guide for comparative case study research is Alexander
Georgeâs âmethod of structured, focused comparisonâ (George, 1979; George &
Bennett, in press; George & McKeown, 1985; George & Smoke, 1974).11 Accord-
ing to George (1979), the comparison is âfocused because it deals selectively with
only certain aspects of the historical case . . . and structured because it employs
general questions to guide the data collection analysis in that historical caseâ
(pp. 61â62; emphasis added). George and his colleagues have outlined several
tasks for researchers who choose the structured, focused case comparison. These
guidelines overlap significantly with the steps we discuss below, although we
consolidate some tasks, divide others, and integrate other scholarsâ advice for
conducting case study research in our discussion. Our aim is both to incorporate
previous discussions of case study research from other disciplines (e.g., public
administration and sociology) and to offer guidelines for conducting case studies
that are accessible and meaningful for political scientists, psychologists, and
political psychologists.
11 In his recently published collection of âmemos,â Stephen Van Evera (1996) also offered step-by-step
advice for case study research.
Comparative Case Study 377
10. Of course, these steps are fairly standard in research design. They are the
traditional steps researchers follow when designing research using almost any
method, yet they are rarely followed for case study research. The form that these
steps take in comparative case study research differs somewhat from that for other
research methods. For example, how reliability is enhanced in a research design is
different for case studies than for experiments, although the same logic underlies
reliability issues in both types of research. Thus, what follows is a discussion of
how traditional research design issues play out in the comparative case study
method.
Step 1: Identify Specific Research Question for Focused Comparison
The first step in comparative case study research, just as for any other method,
is to formulate the specific research question. What phenomena do you want to
explain? What is the dependent variable? The type of research questions most
appropriate for case studies are âhowâ and âwhyâ questionsâthat is, questions
focusing on the underlying process, on the causal nexus between the independent
variables and the phenomena to be explained. Examples are âHow do voters choose
candidates?,â âWhy do unanimity decision rules produce compromise outcomes
in group decision-making?,â and âHow do leaders manage information inconsis-
tent with their beliefs?â Frequency questions (âHow often do different values of
the dependent variable occur?â) are not appropriate for comparisons of a small
number of cases (Yin, 1994).
At this point, the researcher wants to focus the research on a specific phenome-
non and be clear what class of events the research will address. Focusing the
research question is an essential part of Georgeâs âfocused comparisonâ and is
Lijphartâs (1971) advice for dealing with the âtoo many variablesâ problem. As
Lijphart (1971) argued, researchers comparing case studies âmust avoid the danger
of being overwhelmed by large numbers of variables and, as a result, losing the
possibility of discovering controlled relationshipsâ (p. 690). Developing a focused
research question is the best way to identify the most important variables. The
identification of the class of phenomena is also a crucial part of this first step and
will help the researcher both to identify cases for selection and to answer the
question âWhat is this a case of?â For example, suppose you are interested in the
relationship between leadersâ personalities and decision-making outcomes. It is
important at the outset to focus on what leaders you are interested in. In other words,
does the research question apply to all leaders, or to leaders of particular countries
or types of political systems? Focusing on particular types of leaders would make
the identification of explanatory variables and case selection much more manageable.
378 Kaarbo and Beasley
11. Step 2: Identify Variables From Existing Theory
The next question the researcher asks is âWhat independent or explanatory
variables or conditions have been found or hypothesized to explain the dependent
variable?â This is usually accomplished by an extensive literature review. The
researcher should then specify which aspects of existing theory will be singled out
for assessment, again focusing the research question. A large list of potential
explanatory variables investigated in a small number of cases yields an indetermi-
nate research design, and thus the case study researcher must choose a focused set
of candidate explanatory variables to investigate (King et al., 1994). One advantage
of case study research, however, is the ability to track other variables in the cases
that may not be the primary focused explanatory variable(s), but may be important
later in the interpretation of the relationships discovered. This is not uncommon in
experimental studies, wherein measures are frequently taken for a host of variables
whether or not they are of central concern to the study.
The researcher should also make explicit which of the variables to be investi-
gated are hypothesized to be most important for explaining the phenomenon. In
other words, does the theory suggest any ordering of the possible independent
variables? Are any conditions hypothesized to be necessary or sufficient? Hypothe-
sized intervening relationships should also be specified at this point. Although
many research methods are aided by this a priori inventory of relationships among
variables, the case study comparison requires it because this will drive the next step
of case selection.
Step 3: Case Selection
The most important (and difficult) stage in comparative case study research is
the selection of cases. Cases can, of course, be drawn randomly from a population
to eliminate the danger of selection bias. Yet drawing a random sample is often not
possible because the universe of cases is not known or is not accessible.12 At the
heart of systematic investigations is the idea of control. To establish the relationship
between two or more variables, it is necessary to minimize variability in other
variables that may affect the investigated relationship. In the experimental method,
this is approximated by dividing subjects into experimental and control groups
through some form of random assignment. With the use of certain statistical
methods, the same type of control can be achieved through the use of partial
correlations. Although here no âassignmentâ of subjects to conditions occurs, the
influence of certain variables on the relationship being examined can be controlled
for statistically. Control for the comparative case study method is achieved through
12 As King et al. (1994) indicated, random selection does not always solve the problem of selection bias.
See Collier and Mahoney (1996) for an excellent and detailed discussion of the issues surrounding
selection bias in qualitative research.
Comparative Case Study 379
12. case selection. There are three important tasks in selecting cases that involve
control: selecting comparable cases, selecting cases that vary on the dependent
variable, and selecting cases across subgroups of the population to address alter-
native explanations.
3a. Comparable cases. Comparability is the heart of any scientific investiga-
tion and thus is not unique to case study research (see Ragin, 1987, p. 1). Compa-
rability is the basis of Przeworski and Teuneâs (1970) âmost similar systems
designâ in which the researcher chooses cases as similar as possible to minimize
the number of explanatory variables.13 Without comparability, of course, the
researcher will not know if the variation seen in the cases is due to the
explanatory variable under consideration or to the other differences between
the cases (Lijphart, 1971). For example, suppose you are interested in how
juries aggregate individual preferences into a group decision and you hypothe-
size that the size of the jury is the best explanation for this process. If your cases
include juries from different cultures, different time periods, and different legal
systems, you do not know whether the differences in the group decision are due
to the explanatory variable (jury size) or to the other differences between the
cases (culture, time period, and legal system). Thus, if you have reason to
believe culture, time period, and legal system affect how juries make decisions,
you would want to choose cases that are comparable on these dimensions in
order to focus on the effect of size.
Comparability depends on the theoretical basis of the study. Cases do not have
to be comparable on dimensions that do not (or are assumed not to) impinge on the
relationship under investigation. Just as experimentalists are generally not con-
cerned about (or at least tolerate) differing weather conditions on the multiple days
that an experiment takes place (a problem not corrected by random assignment of
subjects to groups), comparative case study researchers should not be concerned
about irrelevant differences in their cases. Building on the previous example, if
your theory or previous evidence suggests that jury size affects the group aggrega-
tion process regardless of culture, time period, or political system, these variables
do not have to be considered to make cases comparable. This stands in contrast to
Diesingâs (1971, pp. 187â188) advice to pick cases that are geographically similar
and that have similar histories if the theory does not suggest relevant dimensions
for comparability. The theory should be more capable of distinguishing impor-
tant factors. Cases should not be chosen for comparability on non-theoretically
derived properties.14 Instead, researchers should choose cases in an effort to
13 Although Przeworski and Teuneâs âmost similar systems designâ focuses on choosing comparable
systems, the same principle applies to any level of analysis.
14 A prerequisite for choosing comparable cases is to define what a âcaseâ or the unit of analysis is for
the investigation. Thus, choosing comparative cases directly follows from the first step of focusing
the research question and identifying the class of phenomena that the question addresses. Is a âcaseâ
a case of jury decision-making (for which any case of a jury decision would yield a comparable case),
or is a âcaseâ a case of jury decision-making in Western cultures, in the late 20th century, in
380 Kaarbo and Beasley
13. control for known or suspected alternative causes of the relationship under
investigation.
3b. Cases with variation in the values of the dependent variable. The most
basic rule in case selection is that selection should allow for the possibility of
variation in the values of the dependent variable (King et al., 1994). Without
variation in values of the dependent variable, the researcher cannot make any causal
inference about the phenomenon because the same explanatory variables may be
present in cases in which the phenomenon is absent. This problem led John Stuart
Mill (1843/1974) to caution scholars against this âmethod of agreement,â and
Millâs caution has been echoed by many later writers on the case study (e.g., Achen
& Snidal, 1989; Collier, 1993; Geddes, 1990). Despite such advice, case study
researchers continue to examine cases of deterrence failure (and not cases of
deterrence success), cases of revolution (and not cases in which a revolution did
not occur), and cases of crisis decision-making (and not cases of routine decision-
making). Such studies can eliminate a hypothesized explanation if it is not present
in the cases (Collier, 1995; Collier & Mahoney, 1996), but cannot conclude that a
factor present in the cases is related to the dependent variable.
To ensure the possibility of variation in the values of the dependent variable,
case study researchers have two choices: selecting cases based on variation in the
values of an explanatory variable, and selecting cases based on variation in the
values of the dependent variable (the latter being Millâs âmethod of differenceâ).
The first option, selecting cases based on categories of an explanatory variable,
causes no inference problems and is therefore considered to be the best solution
(King et al., 1994). With this strategy, for example, the researcher interested in the
relationship between stress and the quality of decision-making would choose cases
in which stress was present (or high) and cases in which stress was absent (or low)
and then proceed to investigate how these independent variables related to the
quality of decision-making.
One of the drawbacks to this strategy (and hence one argument for choosing
the second strategy of selecting cases based on variation in the values of the
dependent variable) is that often it is the value of the dependent variable (i.e.,
that a revolution occurred) that is already known to the researcher. If the
researcher already knows the values of the dependent variable, then this is the
same as selecting the cases on both the independent and dependent variables.
This might cause bias in the case selection if the cases are chosen in such a way
that the independent variable matches up with the known value of the dependent
variable in the hypothesized direction (King et al., 1994). Furthermore, re-
searchers are often inherently interested in explaining the dependent variable
common-law systems (for which cases must be comparable on these dimensions)? The definition of
the unit of analysis must come from the researcherâs theoretical focus or from the population to which
the hypothesis applies.
Comparative Case Study 381
14. (e.g., poor decision-making or revolutions); that is what motivates them to do the
study. Choosing cases based on variation in the values of the dependent variable
hastheadvantageofallowingtheresearchertosaysomethingaboutthedependent
variable under investigation upon its completion. If, using the previous example,
casesinwhichstress(theexplanatoryvariable)waspresentandabsentarechosen
and stress turns out to be unrelated to decision-making quality (the dependent
variable), then this is all the research can conclude. If, on the other hand, cases of
good and poordecision-making(the dependentvariable)are chosen,then,even if
stressturnsouttobeunrelated,theresearchercaninvestigateothervariablesinthe
casesthatmightberelatedtodecision-makingquality.15
3c. Choosing cases across population subgroups. In addition to choosing
comparable cases that vary on the values of the dependent variable, case study
researchers should consider choosing cases with an alternative explanation in mind.
In other words, the case selection design may include variation across some
dimension associated with an alternative explanation of the relationship being
investigated. What the researcher attempts to do in this step is to demonstrate that
the relationship holds across different subgroups of a population. This gives the
researcher, upon the conclusion of the study, the opportunity to challenge an
alternative explanation.16
For example, if you are interested in how personality characteristics affect the
maintenance of enemy images, in addition to choosing cases in which an enemy
image was maintained and cases in which an enemy image changed, you might
consciously choose cases before and after the end of the Cold War. This case
selection would ensure that you could say something about the relationship under
investigation, regardless of the time period. If cases only from the Cold War era or
only from the postâCold War period were examined, you would be open to the
criticism that the studyâs results only apply to that particular era. An alternative
theory might suggest that the nature of enemy images is completely different in
these two historical periods of the international systemâthat enemy images, for
example, are much more malleable in the postâCold War world regardless of
personality characteristics. Of course, choosing cases across different subgroups
of a population proliferates the number of cases needed.17
The difference between choosing comparable cases (step 3a) and choosing
cases across population subgroups (step 3c) concerns the expectations about the
15 A final point in favor of choosing cases based on variation in values of the dependent variable is that,
in many cases, researchers hypothesize multiple (often competing) explanatory variables. Thus,
choosing casesbased on the (single)dependentvariable toensurevariationwouldbea simpler formula
for case selection. See Collier and Mahoney (1996) and Van Evera (1996) for additional issues
concerning case selection and variation in the dependent variable.
16 Choosing cases across subgroups associated with an alternative explanation to show the maximum
leverage of the hypothesis is consistent with Przeworski and Teuneâs (1970) âmost different systems
design.â
17 See King et al. (1994), chapter 5, for more specific advice on this issue.
382 Kaarbo and Beasley
15. applicability of the hypothesized relationship. In choosing comparable cases, the
researcher is recognizing the limits of the hypothesis, arguing that it only
applies to cases that are comparable on a certain number of theoretically derived
dimensions. In choosing cases across subgroups, the researcher is purposefully
stretching the investigation of the hypothesis across a variable derived from an
alternative explanation.
Step 4: Operationalize Variables and Construct a Case Codebook
The fourth major task in designing a comparative case study, and one that is
often overlooked, is to operationalize the variables under investigation. The re-
searcher is essentially constructing a case âcodebookâ to guide the collection of
evidence for the variables in the study. Indeed, King et al. (1994) argued that such
recording and reporting the way the data were collected is even more important in
case studies than is case selection.
George (1979) conceived of the variables in the form of structured questions
that the researcher asks in each case:
Using a standardized set of questions in the controlled comparison is
necessary to assure acquisition of comparable data from the several cases.
In this way, the method of structured, focused comparison will avoid the
all-too-familiar and disappointing experience of traditional, intensive
single case studies in the past which, even when they were instances of a
single class of events, were not performed in a comparable way and hence
did not contribute to an orderly, cumulative development of knowledge
and theory about the phenomenon in question. Instead, as conducted in
the past, each case study tended to go its own way, reflecting the special
interests of each investigator and often, somewhat opportunistically,
being guided by the readily available historical data rather than by a well
defined theoretical focus. As a result, the idiosyncratic features of each
case tended to shape the research questions differently. (p. 62)
Georgeâs structured questionsâthe âstructuredâ element of his âstructured, fo-
cused comparisonââare similar to what Yin (1994, p. 70) called a âcase study
protocol,â a list of case study questions and identification of potential sources of
evidence for answers.
Although we enthusiastically endorse Georgeâs advice to investigate the same
variables across the set of cases in the form of structured questions, we argue that
the case study researcher needs to go beyond structuring questions and to structure
answers as well. In other words, the same set of general categories of answers, or
values of each variable, should guide the researcher across the cases. The key is to
specify in advance what is necessary for the case analyst to see in the evidence to
code a variable in a particular way, to make a judgment of its value. This allows
for enhanced intracoder reliability (the case study analyst uses the same set of
Comparative Case Study 383
16. criteria to assign variables particular values) and provides a basis for intercoder
reliability (the same set of criteria to assign variables particular values can be
used by different analysts). In many case studies, of course, variables will be
qualitatively judged on the basis of subjective interpretations of transcripts, diaries,
etc., but even in this way, a priori categories of the possible variable values can
increase systematicity and reliability.
Just as in the construction of answers to surveys, the analyst should treat the
operationalizations of the possible values of the variables seriously. Operationali-
zations should be generalâapplicable to the class of phenomena, not just to the
specific case under investigationâso that the case codebook could be applied to
cases not in the study. Just as for any codebook, categories, to the extent possible,
should be exhaustive for all possible answers in all cases. Of course, categories
should be mutually exclusive. If some variables have multiple indicators in their
operational definitions, the researcher should specify a meta-decision rule for
deciding the value of the variable. For example, if there are several indicators for
coding a leader as a crusader personality and there are several indicators for coding
a leader as a pragmatist personality, the codebook should specify how to deal with
mixed evidence when indicators of both personalitytypesare found inthe evidence.
The exact nature of the operationalizations of the variables will, of course,
depend on the data sources to be used in the investigation. One advantage of a
comparative case study research design is that a variety of types of data and
techniques for extracting data can be used (Yin, 1994). Content analysis, survey
data, archive data, descriptive statistical analysis, focus groups, and interviews are
all compatible, not competitive, with case study research. The use of multiple
sources is one way to enhance the construct validity of operational definitions (Yin,
1994, p. 34).18
In addition to specifying in advance the operationalizations of the variables,
case study researchers can enhance the reliability and validity of case studies in
other ways. Case study researchers can use judgments made for entirely different
purposes by other researchers to avoid having measurements be influenced by the
hypotheses (King et al., 1994). King et al. also advised using qualified experts to
code the variables who do not know the theory being tested.Even non-expertcoders
who are blind to the hypotheses could demonstrate the intercoder reliability of the
variable measurements, although this may be difficult with certain types of
evidence.
Case study researchers may also consider conducting a pilot study: âThe pilot
case is used more formatively, assisting an investigator to develop relevant lines
of questionsâpossibly even providing some conceptual clarification for the re-
search design as wellâ (Yin, 1994, p. 74). Pilot studies may be necessary when
18 Yin (1994) also argued that the use of multiple sources of evidence allows the researcher to develop
âconverging lines of inquiry, [or] a process of triangulationâ of data sources that makes conclusions
from case studies much more convincing (p. 92; emphasis in original).
384 Kaarbo and Beasley
17. operationalizations do not exist from previous studies or are not specified com-
pletely by the theory.19 After looking at a similar case with similar evidence that
will not be included in the main study, the case analyst is often better able to
fine-tune the operationalizations.
The tasks involved in the construction of a codebookâconstructing structured
answers and structured questions, identifying multiple sources of evidence, testing
for intercoder reliability, investigating pilot casesâmay seem to be tedious work
for a small number of cases. Yet this step addresses the most often heard criticism
of the case study: unreliability. Following rules for systematic extraction of data,
and reporting these rules to readers, enhances the credibility of conclusions from
case studies. Unfortunately, this step is rarely taken in case study research (George,
1979).20
Step 5: Code-Write Cases
Once the codebook is constructed and the evidence is accumulated, the
variable should be coded and the results presented. These are actually two separate
tasks. Indeed, Yin (1994) suggested creating a case study database, different and
separate from the report, arguing that âtoo often, the case study data are synony-
mous with the evidence presented in the case study report, and a critical reader has
no recourse if he or she wants to inspect the database that led to the case study
conclusionsâ (p. 95).
In terms of presentation, the narrative remains the most used and preferred
form (Yin, 1994). The narrative case study tells a story based on the variables and
coded values and allows the researcher to show the variables in their context.21 This
is a way to demonstrate the realism and temporal progression that is an advantage
of the case study method.22 In the narrative, any uncertainty on the part of the
analyst should, of course, also be presented (King et al., 1994). The length and
detail of the narratives depends on the research question, the number of variables,
and the number of cases.
Although the narrative form of presentation demonstrates the richness of the
data, it is not the only form available to the case study researcher and indeed can
become impossible when a large number of case studies are involved. Alterna-
tively, the case analyst can present a question-and-answer format or present only
19 This is one of the uses of a case study to explore and refine theory discussed earlier.
20 Analysts are, of course, still free to observe variables not included in their codebook, and even to
question the particular operationalizations that they have chosen. A codebook should not hinder the
valuable insights afforded analysts as they examine phenomena in their naturally occurring context.
These, however, should be noted as such when researchers report their findings.
21 For a discussion of different purposes of case narratives, see Abbott (1992, pp. 62â80).
22 The narrative form also has the advantage of conforming to the way people think about events. As
Pennington and Hastieâs (1986) research demonstrates, people think in terms of stories, not in terms
of comparing the values of variables in isolation.
Comparative Case Study 385
18. coded results in table form with short case descriptions in an appendix (Yin, 1994).
Combinations of narratives and short descriptions might be used, depending on the
results. For example, the researcher might want to demonstrate in more detail the
cases that fit or do not fit the hypotheses the best. In whichever form the case analyst
chooses to present the cases, the presentation of results should include more explicit
discussion of method than is usually found in case studies (George & McKeown,
1985).
Step 6: Comparison and Implications for Theory
After the cases have been coded for the variables, the case study analyst looks
for patterns within and across cases. In the congruence procedure, the analyst
determines whether any independent variables differentiate across the different
outcomes of the dependent variable (George & McKeown, 1985). In other words,
the analyst asks: Does the value of x relate to the value of y in the predicted way?
The analysis stage can follow another form of logic in which a pattern
predicted by the theory is matched against the pattern seen in the cases. In other
words, a number of different explanatory variables must be present and related to
each other in the right form for the hypothesis to be accepted. This technique of
âpattern matchingâ is a key advantage of the case study method (Campbell, 1975;
Collier, 1993). In pattern matching, a single predicted pattern can be compared to
the pattern observed in the cases or, alternatively, mutually exclusive rival patterns
can be compared for their consistency with the pattern observed in the cases (Yin,
1994, p. 108). This essentially increases the number of observations made within
each case study. In recanting his earlier âstrong rejectionâ of the case study method,
the well-known psychologist Donald Campbell (1975) applied the statistical con-
cept of degrees of freedom to case studies. According to Campbell, theories usually
have a number of predictions that the social scientist uses to âpattern matchâ against
a single case. In other words, degrees of freedom in a case study means âtestingâ
multiple implications of the theory.
When the pattern that is investigated is temporal in character, the case analyst
engages in what George and McKeown (1985) termed âprocess tracingâ or inves-
tigating a âdecision process by which various initial conditions are translated into
outcomesâ (p. 35).23 As Yin (1994) argued, âthe ability to trace changes over time
is a major strength of case studiesâwhich are not limited to cross-sectional or static
assessments of a particular situationâ (p. 113). However, the data requirements for
process tracing are considerably greater than for the simple congruence procedure
and thus may not be practical for some case study research.
23 For George and McKeown (1985), process tracing means more than the analysis of temporal patterns
within cases; it also involves interpretivism in that it is âan attempt to reconstruct actorsâ definitions
of the situationâ (p. 35).
386 Kaarbo and Beasley
19. Examining the multiple implications of the theory or engaging in process
tracing within a single case study also results in certain limitations that should be
made clear by the case analyst. Much like choosing comparable cases by minimiz-
ing variation on alternative explanatory variables, examining patterns and pro-
cesses through time within a single case study does not allow for any between-cases
variation. Although a particular observed pattern may or may not conform to
theoretical predictions, it is difficult within a single case study to determine whether
a particular set of observations is idiosyncratic to that case or not. Further, a
particular pattern of observations within a single case study could be due to the
impact of a causal factor that cumulates through time but which is thought to be
independent of the causal process revealed in the case. For example, a particular
pattern of activity of presidential initiatives to secure approval of legislation may
reflect a predicted pattern of intensification through time due to a particular
personality trait, but for a given president regarding a particular initiative it may
also reflect a gradual accumulation of initial conditions such as public opinion,
economic prosperity, or other nonâpersonality-based factors. Of course, investi-
gating more than one case study at a time helps to address some of these concerns,
as patterns and processes noted within one case study can then be systematically
compared to the other case studies that have been comparably performed.
Once the initial assessment of the theoretical implications is made, the case
study analyst may want to âreturnâ to the cases to investigate candidate explana-
tions for the relationships discovered. In other words, certain interpretations arising
from the initial conclusions can often be discounted or suggested by analyzing the
âextraâ material included in the case studies. Although the researcher may need to
design a new research plan and select new cases to better âtestâ these implications,
case study research can offer the researcher more material on which to build
speculative interpretations of results. This is a heuristic value built into every case
study research design.
Conclusions: Why Should Political Psychologists Care About
the Case Study Method?
It is our hope that this examination of the case study method will help to guide
researchers in their use of this methodology and will begin to dispel several
stereotypical conceptions of the case study method. As we suggested at the outset,
the case study has been plagued by various misconceptions, attributable more to
the way it has been used than to its actual potential. We have suggested that,
although valuable as a tool for understanding idiosyncratic historical events, the
method is also valuable as a tool for generating and exploring theories and testing
hypotheses. Further, we have offered various techniques for dealing with the issue
of the ratio of observations to hypotheses. We have also offered, by way of a review
of general design procedures, a set of steps for gaining some measure of control in
the comparative case study method. Although by definition our conception of the
Comparative Case Study 387
20. case study method disavows direct manipulation of either context or phenomenon,
case selection procedures do allow a degree of discrimination in determining
relationships between and among variables that is similar to that afforded by
experimental manipulation or statistical control. Finally, the resounding critique of
subjectivity on the part of the researcher has been addressed by suggesting
procedures that can significantly improve the systematicity and reliability of the
case study method.
Beyond the general value of case study research, the case study method is
particularly suitable for political psychology. We base this proposition on Margaret
Hermannâs (1986) identification of the tenets of the field of political psychology.
In her answer to the question âWhat is political psychology?â Hermann proposed
five common themes that characterize research in this area:
1. Focus Is on the Interaction of Political and Psychological Phenomena
2. Research Is Responsive and Relevant to Societal Problems
3. Context Can Make a Difference
4. Emphasis Is on Process as Well as Outcome
5. There Is a Tolerance of Multiple Methods for Gathering Data
Comparative case study research, we believe, is particularly relevant to the last
three of Hermannâs tenets.24
First, case study research is relevant to the tenet that context can make a
difference. Hermann (1986) wrote that âwithin political psychology there is a
healthy dialogue and debateâ on the importance of contextual factors such as time,
culture, and political structures, but that âthere is a growing recognition that such
contextual factors help not only to shape what an individual is like, but also to limit
what an individual can do politicallyâ (p. 3). Case study research internalizes this
tenet, as âcontextâ is part of the definition of the case study offered earlier in this
paper. Case study research means that a real-world phenomenon is investigated
within its naturally occurring context, with the recognition that context can make
a difference.
Second, the case study method is consistent with the tenet that the process is
as important as the outcome. Hermann (1986) argued that
psychological and political phenomena often seem to have their most
direct impact on one another in determining political processesas opposed
24 This is not to suggest that the case study method is not relevant to the first two tenets. Case study
research is certainly appropriate for the study of the interaction of political and psychological
phenomena and for research that is responsive and relevant to societal problems. Indeed, our
discussion in this paper of case studies whose goal is intervention and the desirability of choosing
cases based on variation in the values of an interesting dependent variable reflects this second tenet.
Our point in this section, however, is that there is something about the case study method in particular
that is consistent with the last three of Hermannâs tenets.
388 Kaarbo and Beasley
21. to political outcomes. Once delineated in one political setting, descrip-
tions of processes offer the possibility of generalization to other political
contexts with somewhat similar characteristics. By ascertaining what
some of these cross-cutting processes are, political psychologists can
begin to build more general explanations and theories. (p. 3)
Case study research is often directed at identifying the underlying process. As
discussed above, the most appropriate research questions for case study research
are âhowâ and âwhyâ questions. Furthermore, pattern matching (especially tem-
poral pattern matching or process tracing), as a technique for analysis in case study
research, emphasizes the process by which initial conditions are translated into
outcomes. In this sense, process is central to most case study research.
Finally, case study research is compatible with political psychologyâs toler-
ance of multiple methods for gathering data. As Hermann (1986) wrote, âpolitical
psychology appears not to be wedded to one type of methodology. Its practitioners
employ a variety of methodsâ (p. 3). Not only can the case study method serve to
complement other research methods, it can also be multimethodological in itself,
as a number of techniques (e.g., content analysis, archival analysis, interviews) can
be incorporated into a case study research design.25
Clearly, then, the case study method can be a valuable research tool and is
particularly suited to the needs and interests of those studying within the field of
political psychology. In this light, it is unfortunate that some of the misconceptions
about case studies have contributed to the methodological barrier between political
science and psychology. It has not been our argument here that the case study
should supplant other methods, nor have we tried to systematically articulate each
advantage and disadvantage associated with case study research relative to other
methods. Rather, we hope to have provided a practical guide to doing case study
research and in the process helped to dispel certain misconceptions and thereby
further the development of a common language for political psychology.
ACKNOWLEDGMENTS
Earlier versions of this paper were presented as part of the methods workshop
series at the Summer Institute of Political Psychology at The Ohio State University,
1992â1997. We would like to thank the participants in those workshops, as their
questions and comments helped to refine our thinking about this topic over the
years. We also thank Alexander George as well as the editors and anonymous
reviewers for suggestions on earlier drafts. Correspondence concerning this article
should be addressed to Juliet Kaarbo, University of Kansas, Department of Political
Science, 504 Blake, Lawrence, KS 66045. E-mail: kaarbo@falcon.cc.ukans.edu
25 See Tetlock (1983) on the advantages of multimethod convergence in political psychology.
Comparative Case Study 389
22. REFERENCES
Abbott, A. (1992). What do cases do? Some notes on activity in sociological analysis. In C. C. Ragin
& H. S. Becker (Eds.), What is a case?: Exploring the foundations of social inquiry (pp. 53â82).
Cambridge: Cambridge University Press.
Achen, C.,&Snidal, D. (1989). Rational deterrence theoryand comparativecase studies.World Politics,
41, 143â169.
Baker, T. L. (1988). Doing social research. New York: McGraw-Hill.
Campbell, D. T. (1975). Degrees of freedom and the case study. Comparative Political Studies, 8,
178â193.
Collier, D. (1993). The comparative method. In A. W. Finifter (Ed.), Political science: The state of the
discipline II (pp. 105â119). Washington, DC: American Political Science Association.
Collier, D. (1995). Translating quantitative methods for qualitative researchers: The case of selection
bias. American Political Science Review, 89, 461â466.
Collier, D., & Mahoney, J. (1996). Insights and pitfalls: Selection bias in qualitative research. World
Politics, 49, 56â91.
Diesing, P. (1971). Patterns of discovery in the social sciences. New York: Aldine.
Eckstein, H. (1975). Case study and theory in political science. In F. I. Greenstein & N. W. Polsby
(Eds.), Handbook of political science (pp. 79â138). Reading, MA: Addison-Wesley.
Farnham, B. (1992). Roosevelt and the Munich crisis: Insights from prospect theory. Political
Psychology, 13, 205â235.
Fife-Schaw, C., & Breakwell, G. M. (1990). Predicting the intention not to vote in late teenage: A U.K.
study of 17- and 18-year-olds. Political Psychology, 11, 739â755.
Fishel, J. (1992). Leadership for social change: John Vasconcellos (D-CA) and the promise of
humanistic psychology in public life. Political Psychology, 13, 663â692.
Geddes, B. (1990). How the cases you choose affect the answers you get: Selection bias in comparative
politics. In J. A. Stimson (Ed.), Political analysis (vol. 2, pp. 131â150). Ann Arbor, MI: University
of Michigan Press.
George, A. L. (1979). Case studies and theory development: The method of structured, focused
comparison. In P. G. Lauren (Ed.), Diplomacy: New approaches in history, theory, and policy
(pp. 43â68). New York: Free Press.
George, A. L., & Bennett, A. (in press). Case studies and theory development. BCSJA Security Studies
Series. Cambridge, MA: MIT Press.
George, A. L., & McKeown, T. J. (1985). Case studies and theories of organizational decision making.
In R. F. Coulam & R. A. Smith (Eds.), Advances in information processing in organizations (vol.
2, pp. 21â58). Greenwich, CT: JAI Press.
George, A. L., & Smoke, R. (1974). Deterrence in American foreign policy. New York: Columbia
University Press.
Gilbert, R. E. (1995). The political effects of presidential illness: The case of Lyndon B. Johnson.
Political Psychology, 16, 761â776.
Goldmann, K. (1988). Stability and change in foreign policy. Princeton, NJ: Princeton University Press.
Groth, A. J., & Britton, S. (1993). Gorbachev and Lenin: The psychological walls of the Soviet âgarrison
state.â Political Psychology, 14, 627â650.
Hermann, M. G. (1986). What is political psychology? In M. G. Hermann (Ed.), Political psychology
(pp. 1â10). San Francisco: Jossey-Bass.
Hermann, M. G. (1989, August). Political psychology: Fad, fantasy, or field. Paper presented at the
annual meeting of the American Psychological Association, New Orleans.
Isaak, A. C. (1985). Scope and methods of political science. Chicago: Dorsey.
390 Kaarbo and Beasley
23. Johnson, J. B., & Joslyn, R. A. (1991). Political science research methods. Washington, DC:
Congressional Quarterly Press.
King, G., Keohane, R. O., & Verba, S. (1994). Designing social inquiry: Scientific inference in
qualitative research. Princeton, NJ: Princeton University Press.
Lijphart, A. (1971). Comparative politics and the comparative method. American Political Science
Review, 65, 682â693.
Lijphart, A. (1975). The comparable-case strategy in comparative research. Comparative Political
Studies, 8, 158â177.
Milburn, M. A., & McGrail, A. B. (1992). The dramatic presentation of news and its effects on cognitive
complexity. Political Psychology, 13, 613â632.
Mill, J. S. (1974). A system of logic. Toronto: University of Toronto Press. (Original work published
1843)
Orum, A. M., Feagin, J. R., & Sjoberg, G. (1991). Introduction: The nature of the case study. In J. R.
Feagin, A. M. Orum, & G. Sjoberg (Eds.), A case for the case study (pp. 1â26). Chapel Hill, NC:
University of North Carolina Press.
Pennington, N., & Hastie, R. (1986). Evidence evaluation in complex decision making. Journal of
Personality and Social Psychology, 51, 242â258.
Przeworski, A., & Teune, H. (1970). The logic of comparative social inquiry. New York: Wiley.
Ragin, C. C. (1987). The comparative method: Moving beyond qualitative and quantitative strategies.
Berkeley, CA: University of California Press.
Ragin, C. C., & Becker, H. S. (Eds.) (1992). What is a case?: Exploring the foundationsof social inquiry.
Cambridge: Cambridge University Press.
Raven, B. H. (1990). Political applications of the psychology of interpersonal influence and social
power. Political Psychology, 11, 493â520.
Steinberg, B. S. (1991). Shame and humiliation in the Cuban missile crisis: A psychoanalytic
perspective. Political Psychology, 12, 653â690.
Suedfeld, P. (1992). Bilateral relations between countries and the complexity of newspaper editorials.
Political Psychology, 13, 601â612.
Tetlock, P. E. (1983). Psychological research on foreign policy: A methodological review. Review of
Personality and Social Psychology, 4, 45â78.
Tiedens, L. Z. (1997). Optimism and revolt of the oppressed: A comparison of two Polish Jewish ghettos
of World War II. Political Psychology, 18, 45â69.
Van Evera, S. (1996). Guide to methodology for students of political science. Cambridge, MA: Defense
and Arms Control Studies Program, Massachusetts Institute of Technology.
Yin, R. (1994). Case study research. Beverly Hills, CA: Sage.
Comparative Case Study 391