Case selection How to test a sufficient or a necessary condition with a case study

For testing whether A is a necessary condition for B, there are also two possibilities for selecting the case: ■ selection on the basis of the presence of the dependent concept “If there is B, then it is tested if there is also A”; ■ selection on the basis of the absence of the independent concept “If there is no A, then it is tested if there is also no B” non-A is a sufficient condition for non-B. Depending on the objective of the research a “most likely” or a “least likely” case can be selected. A “most likely” case is an instance of the object of study in which confirmation of the hypothesis is likely. Such selection strategy can be used when the proposition is tested for the first time “initial theory-testing research”. This strategy is also pos- sible when the outcomes of earlier tests result in doubts about the sup- port for the proposition in the domain where it was tested. The researcher may then want to select a “most likely case” to find parts in the domain where the proposition could be supported. A “least likely” case is an instance of the object of study in which support for the proposition is not likely. This strategy may be used when the outcomes of earlier tests indicate support for the proposition, and the researcher wants to know what the boundaries of the domain are. This is import- ant for determining the generalizability of the theory. For case selection, as it is described here, it is necessary to measure the value of the independent or dependent concept before the actual test is conducted, i.e. before “measurement” has occurred. However, it might not be feasible to measure the value of the relevant concept at this stage of the research. In this case, an alternative strategy for case selection is that a candidate case is selected and that it is verified in the measurement phase of the research whether the concept is indeed present. If not, then the case cannot be used for testing the hypothesis and another case must be selected.

5.1.4 Hypothesis

A proposition is a statement about a relation between concepts. For testing, a proposition must be reformulated into a hypothesis. A hypothesis is a statement about a relation between variables in which the variable is a measurable indicator of the concept. In this type of theory-testing the hypothesis can be formulated quite easily. If the proposition specifies a sufficient condition and a case is selected in which the condition is present, the hypothesis is that the effect is also present in that case. If a case is selected in which the effect is absent, the hypothesis is that the condition is also absent in that case. If the proposition specifies a necessary condition and a case is selected in which the effect is present, the hypothesis is that the condition is also present in that case. If a case in which the condition is absent is selected, the hypothesis is that the effect is also absent in that case.

5.1.5 Measurement

In order to compare the prediction expressed in the hypothesis with the facts of the case, these facts must first be measured. Measurement is a process in which a score or scores are generated for analysis. Measurement consists of a data collection, and b coding. Measure- ment issues are discussed in Appendix 1 “Measurement”. As mentioned above, a complication regarding the case selection in this specific type of theory-testing case study is that the value of one of the concepts must be known before case selection. Otherwise it is not possible to identify and select this specific case in the first place. Hence, the principles of measurement as discussed in Appendix 1 also apply to the procedures of case selection.

5.1.6 Data presentation

For testing a sufficient condition it must first be shown that the condition A was present or effect B was absent in the case, so that the case can be accepted for the test. Next, the observed score of effect B or the score of condition A must be present. For testing a necessary condition it must first be shown that the effect B was present or the condition A was absent in the case, so that the case can be accepted for the test. Next, the observed score of condition A or the score of effect B must be present. In a serial or parallel sin- gle case study, the data must be presented for each case separately.

5.1.7 Data analysis

Data analysis is the interpretation of scores obtained in a study in order to generate the outcome of the study. After having measured the actual score of either effect B or condition A, data analysis consists of testing the hypothesis. Hypothesis-testing is comparing the observed pattern