A case study is a research approach in which one or a few instances of a phenomenon are studied in depth.
Case studies were the predominant research approach at the beginning of modern social science. This is reflected, for example, in the work of the Austrian- born anthropologist Bronislaw Malinowski and the Chicago School of sociology, both of which embraced case study research. Nevertheless, after World War II, quantitative methods gained a hegemonic position, at least among methodologists. It is noteworthy that even during this heyday of quantitative research many important studies that provided theoretical break- throughs and have entered the pantheon of classic works, such as Graham Allison’s study on the Cuban missile crisis in 1971, were based on the case study approach. During recent years, we have seen not only a resurgence of case studies in most disciplines but also unprecedented methodological reflection on this approach. This can be seen as an alignment of epistemology/methodology to ontology/theory. The strong emphasis in recent theoretical approaches of aspects such as “ideas” and “timing” is favorable for case study approaches. Social constructivist theories stress the importance of individual perceptions or hegemonic discourses in social processes. Case studies are much better suited than large-N studies for tracing these ideas because they can invest heav- ily in in-depth interviews or discourse analysis.
Game theory and theoretical notions such as “path dependency” stress the importance of timing for explaining specific outcomes. Again, producing a detailed historical account is certainly one of the major strengths of case studies.
This entry first discusses the nature of case studies, their advantages and disadvantages, and three perspec- tives on their use. The final sections of the entry are devoted to the very important steps in doing case study research: case selection and data analysis. Although the praxis of doing case study research is dominated by the challenges of collecting empirical evidence, this stage is not discussed here because useful information for deal- ing with these challenges can be found in other entries.
What Is a Case Study?
There is no consensus on the basic characteristics of case studies. One reason for this is the fact that the
term is not restricted to social science research but rather is used in many practical contexts. Therefore, the understanding of case studies extends from being a specified tool in a purely positivist scientific research endeavor to being a pedagogical strategy in education and social learning processes. Qualitative case study researchers argue that cases must be seen as configurational context- and/or path-dependent entities. They advocate in-depth strategies such as
“thick description” and “process tracing,” and they opt for a “case-centered” approach rather than the
“variable-centered” one that dominates in quantitative/
positivist research.
Case studies focus on one or a few instances, phe- nomena, or units of analysis, but they are not restricted to one observation. Nevertheless, the bound- aries are not fully clear. On the one hand, John Gerring would exclude case studies that lack any spa- tial or temporal variation. On the other hand, Charles Ragin’s qualitative comparative analysis (QCA) meth- ods try to expand the reach of case-centered research approaches beyond the usual limits toward the range of 10 to 60 cases. However, the particular strength of qualitative case study research—the ability to study the case in depth, which is the best-known aspect of research on a captive population—may be lost in this endeavor to bridge the quantitative–qualitative gulf.
Advantages and
Disadvantages of Case Studies To understand the specificities of case study research, it is useful to compare it with the two other main research approaches: experiments and large-N sur- veys. Such comparisons reveal that the main differ- ence between case studies and experiments is that in experiments cases are created by the researcher and factors of influence can be controlled.
The relationship between case studies and large-N studies lies in the specific affinities and comparative advantages of these two approaches with respect to specific goals and contexts. First, it is broadly accepted that case studies have been the major source of theoret- ical innovation, whereas large-N studies have their strength in controlling the empirical scope of new theoretical concepts. Second, whereas large-Nstudies tend to focus on causal research goals, case study research has an affinity toward descriptive goals. This does not mean that case study research is not con- cerned with causal questions, but it usually takes the 68———Case Study
descriptive–interpretive elements more seriously. In addition, case studies are often concerned with pinning down the specific mechanisms and pathways between causes and effects rather than revealing the average strength of a factor that causes an effect. Third, even positivist methodologists accept that case studies have a strong comparative advantage with respect to the
“depth” of the analysis, where depth can be understood as empirical completeness and natural wholeness or as conceptual richness and theoretical consistency. In contrast, large-Nstudies have advantages in terms of the “breadth” of the propositions, an important argu- ment in contexts where there are many similar cases or where a homogeneous population of cases is assumed.
Fourth, large-Nstudies are better equipped for secur- ing external validity by using statistical means of con- trol. In contrast, case studies have advantages with respect to construct and internal validity. The argument for better construct validity is based on the fact that case studies can use more and more diverse indicators for representing a theoretical concept and for securing the internal validity of causal inferences and/or theo- retical interpretations for these cases.
Three Different Views on Case Studies As the methodological reflection on case studies unfolds, it is increasingly obvious that there are quite different understandings of case study research. We can distinguish among three ideal types: naturalism, positivism, and constructivism.
With respect to main goals, naturalists want to gen- erate practical and detailed knowledge, positivists aim at the establishment of conceptually rather narrow but law-like propositions and models that allow predic- tions, and constructivists see the empirical endeavor of doing case studies as a contribution and check to a theoretical discourse.
Naturalists advocate “natural generalization”
through social diffusion and learning processes.
Researchers themselves do not try to generalize beyond the case under investigation, but the findings can be taken up by others if they perceive a “fit” to their cases. The generalizations made by positivists can be labeled “statistic generalizations” based on drawing logical inferences from a sample of cases to a specified population. The third approach to generalization was called “analytic generalization” by Robert Yin, but a more distinctive label is “theoretical generalization”
because it is characterized by drawing interpretive
inferences from a variety of observable objects to meaningful abstract concepts. For example, Graham Allison and Philip Zelikow used the case study on the Cuban missile crises to show how an “organizational behavior” model and a “governmental politics” model shed additional light on foreign policy decision mak- ing, in contrast to a unified/rational actor model.
Both naturalists and positivists make the ontologi- cal assumption that there exists a single objective reality that is independent of human observation.
Naturalists try to reveal the authentic nature of a social phenomenon or the detailed elements of a causal process by getting as close as possible. Therefore, strategies such as participatory observation and the use of empathy are fully accepted. The positivists opt for “control” instead of “closeness” to reveal an objective reality. The methodological emphasis is not on bridging the gulf between reality and researcher but rather on revealing the relationship between the particular (the individual case) and the universal (the population). Constructivists, in contrast, do not assume any single reality and believe that empirical reality and theoretical concepts are mutually constitutive. For them, bridging is focused on narrowing the gap between concrete observations and abstract meanings using interpretive techniques. Because interpretation loses much of its associative quality if it is pressed into quantitative methods, constructivists adopt another means of control. They use a plurality of the- ories to understand and analyze cases.
Choosing Cases
For naturalists, it is the intrinsic interest in a specific case that motivates case study researchers. A case may have an important real-life impact, and therefore it makes sense to concentrate their scholarly efforts on the internal complexity of the case, leaving aside any prior considerations about potential generalizations.
Having access is another important criterion for selecting cases from this perspective.
Careful selection of cases is essential for positivists because it enables them to draw statistical generaliza- tions and because they accept only covariation as a basis for causal inference. These two aspects are con- nected to corresponding selection criteria. First, a case can be selected because it has a specific position within the larger population. Prior quantitative studies are nec- essary to reveal typical/representative, diverse, extreme, deviant, and (statistically) influential cases. Selecting a Case Study———69
case in such embedded case studies can be used to test or differentiate the causal pathways for a statistically proven causal proposition. Second, a few cases can be selected on the basis of their similarity or difference, making it possible to draw causal inferences through cross-case comparisons. Such attempts to control vari- ance and to select what Arend Lijphart called “compa- rable cases” either with reference to John Stuart Mill’s
“method of agreement” and “method of difference” or with reference to Adam Przeworski and Henry Teune’s
“most similar systems design” and “most different sys- tems design” are quite common, although they do not hold up to rigorous logical standards.
Constructivists opt for selecting theoretically “cru- cial cases.” Harry Eckstein’s notions of “least likely”
and “most likely” cases have been taken up as devices for theory-oriented selection of cases. Constructivists do not share the positivists’ covariational interpreta- tion of crucial cases but do share their conviction that the selection of crucial cases makes (theoretical) gen- eralizations possible. Most likely cases are cases where a plurality of diverse indicators of an internally coherent theory would make it very likely that another empirical aspect (e.g., an indicator of a dependent variable but also a causal process) also corresponds to the logic of that theory but does not. The famous studies of Malinowski (in 1926), William Foote Whyte (in 1943), and Lijphart (in 1975) followed this logic, and all of them seriously undermined the hege- mony of theories that dominated their fields at the time. The selection of least likely cases is aiming at what has been labeled the “Sinatra inference”: If a theory can make it here, it can make it everywhere. A paradigmatic example is the study by Robert Michels (in 1962) on oligarchies in organizations.
Positivists and constructivists need to invest a lot before strategically selecting useful cases to reach their specific kinds of generalization. Nevertheless, if they have selected their cases on the basis of more practical reasons, they should still reflect on the posi- tion of their cases with respect to populations or theo- ries if they wish to reach their goals.
Describing, Analyzing, Interpreting, and Documenting Empirical Evidence For naturalists, it is especially important to provide a comprehensive and consistent picture of a case. Their inductive approach is associated with thick descriptions, narratives, and process tracing as adequate means to
analyze and document the evidence. Alexander George and Andrew Bennett defined process tracing as a method that attempts to identify the intervening causal processes—the causal chain and causal mechanism—
between an independent variable (or variables) and the outcome of the dependent variable. Because they insisted that every element of the causal chain must be empiri- cally documented and stated an affinity to a scientific realist understanding of causal mechanisms, the empha- sis is on chronological narratives. In a second step, a detailed narrative might be combined with abstract con- cepts that provide general assumptions about the func- tioning of time (e.g., concept of path dependence) to make the conclusion more plausible. The best-known example for path dependency is the explanation of the widespread adoption of the QWERTY keyboard through positive feedback loops based on an initial large market share and on network effects (and not on intrinsic supe- riority). Nevertheless, the proof of a causal relationship at a specific sequence (e.g., at a “critical juncture”) relies heavily on finding and documenting a “smoking gun,”
that is, evidence for a causal connection that common sense would not doubt.
Positivists instead employ quasi-experimental methods to draw conclusions from empirical data. The most important means are cross-case comparisons, splitting the case into multiple entities, temporal seq- uencing, and counterfactual thought experiments as a means to allow “controlled” spatial, longitudinal, and/or imaginary covariational comparisons and to draw logical conclusions.
The analytic approach that corresponds to a con- structivist account is an extended and specified under- standing of what George and Bennett called the congruence method. In contrast to naturalistic approaches, it has a strong deductive element because it begins with theories and assesses their comparative strength in understanding and explaining empirical cases. A constructivist would not limit this method to comparing the theoretical expectations with the empirical reality on a variety of indicators for the dependent and independent variables. Instead, search- ing for (non)congruence is extended to causal pro- cesses. In contrast to the naturalist’s inductive understanding of process tracing and the scientific realist’s account of causal mechanisms, the construc- tivist deduces empirical implications that correspond to a specific theory all along the way from the causal factors to the causal processes to the effects. Only quite general theories that embody a fundamental 70———Case Study
causal mechanism, usually some kind of microfoun- dation, are capable of serving as a basis for this kind of deduction. This ensures that case studies become embedded in the fundamental theoretical debates within the social sciences. The quality of a case study, thus, does not depend on providing detailed evidence for every step of a causal chain; rather, it depends on a skillful use of empirical evidence for making a con- vincing argument within a scholarly discourse that consists of competing or complementary theories.
The adequate structure for documenting case study findings is chronological for naturalists, linear–analytic for positivists, and comparative for constructivists.
Joachim K. Blatter
See alsoConstructivism; Generalizability; Historical Research; Interpretive Research; Narrative Analysis;
Naturalistic Inquiry; Positivism
Further Readings
George, A., & Bennett, A. (2005).Case study and theory development in the social sciences.Cambridge: MIT Press.
Gerring, J. (2007).Case study research: Principles and practices.Cambridge, UK: Cambridge University Press.
Gomm, R., Hammersley, M., & Foster, P. (Eds.). (2000).
Case study method: Key issues, key texts.London: Sage.
Stake, R. (1995).The art of case study research.Thousand Oaks, CA: Sage.
Yin, R. (2003).Case study research: Design and methods (3rd ed.). Thousand Oaks, CA: Sage.