 |
Case selection. Box 21. 2 descriptive case study and a causal. Strengths and limits of case study research:what do you want to know?
|
|
|
|
CASE SELECTION
Gerring (2017: Chapter 3) addresses a critical question for case study research: how do we decide which cases to select for intensive study? Like others, he identifies two broad types of case study; these, in turn, relate to the goals of a case study: to describe a case or handful of cases (descriptive) or to explain an outcome (causal). In addition, he identifies a third category, which is applicable to both descriptive and causal goals, in which case selection is guided by practical concerns such as ‘intrinsic importance’ or whether a researcher has access to relevant evidence (Gerring, 2017: 42).
For a descriptive case study, case selection is of a typical case, or of two or more sub-types of cases (diverse cases). For a causal case study, case selection will depend
on which type of causal case study you intend to carry out: exploratory (to generate hypotheses), estimating (to estimate outcomes) or diagnostic (to determine interrela- tionships among the variables). In Box 21. 2, I describe two types of exploratory causal studies. These seek to generate hypotheses on the relationship between one or more variables and an outcome. Common types of exploratory causal studies are most-similar and most-different. I give examples with reference to our research on plagiarism in Box 21. 2.
BOX 21. 2 DESCRIPTIVE CASE STUDY AND A CAUSAL
(EXPLORATORY) CASE STUDY
Let’s assume we have carried out a cross-case study of plagiarism. Using the survey findings, we decide to carry out a descriptive case study and a causal (exploratory) case study. For the former, we identify six student sub-groups, arrayed as a 3 x 2 table, which cross-tabulates their behaviour (never plagiarized, sometimes plagiarized and often plagiarized) and the detection of their behaviour (no, not detected; yes, detected). We then select six students who are typical of that cell for careful examination. This would be a descriptive study of sub-types (diverse).
In the causal case study, we want to understand a surprising finding from our survey: among the subset of students with a high degree of risk-taking behaviour, some of which was illegal, one group was never caught for attempted plagiarism, while another group was. Case selection would be based on a most-similar design, which means the case is the same on a background factor that may cause the outcome (risk-taking), but different on the outcome. We decide to select four cases for intensive examination: one male and female each for the two groups. We seek to understand what the mechanisms are that facilitate the ability of one group of students who are similar on risk-taking to evade (or not) detection for plagiarism. For a most-different design, we select four cases which are the same on the outcome, but different on the background factor that we think should have caused the outcome.
There are many considerations in selecting cases, but, ultimately, how you decide to select depends on the goals of your research. Case selection should be done with awareness of the relevant rule you are using to select cases, and this, in turn, should be explained in your research. Unfortunately, many researchers do not know the rules and do not report them. Imagine that a survey researcher did not explain the decisions that were made in drawing a sample of cases. This would be unheard of. Thus, case study researchers need to become sophisticated in their knowledge of case selection and how it is reported.
|
|
|
STRENGTHS AND LIMITS OF CASE STUDY RESEARCH: WHAT DO YOU WANT TO KNOW?
All methods of conducting research have strengths and weaknesses. Let us consider what these are by comparing case study and cross-case research. The comparison is not absolute, but reveals ‘methodological affinities’ and a set of ‘trade-offs’ (Gerring, 2007: 38, emphasis in original; see Chapter 3). In general, case study research is recommended if you seek to:
· generate hypotheses, rather than test hypotheses
· have internal validity (ensure that the interrelationship(s) in your sample are cor- rect) rather than external validity (the ability to generalize from a sample to a larger population)
· understand causal mechanisms (know how one or more variables relate to the others) rather than predict causal effects
· have a deep, rather than a broad, understanding of phenomena.
Internal validity is typically more relevant for causal than descriptive case studies, and causal mechanisms are more relevant for causal than descriptive case studies.
To understand these terms, let’s turn to the second case study (causal exploratory) in our plagiarism research. We are interested to generate hypotheses on the surprising finding from the survey of students. We want to be sure we have identified all the relevant variables that are associated with high risk-taking, but with different out- comes in plagiarism detection (internal validity). We want to show how the variables are causally related (causal mechanisms), and we are able to gain a deep understand- ing of the process. After we complete the case study, and if our survey data permit it, we can test hypotheses generated from the case study. With a large sample size, we will feel more confident in being able to generalize our findings to a larger popula- tion (external validity) and to have a broad understanding of how risk-taking behaviour relates to plagiarism detection for male and female students.
|
|
|
