• Choose a category
• All
• Classic-Fiction

QCA allows for the possibility of equifinality. It can also use interval data by adding nominal categories to represent interval values (though this adds to the complexity of the analysis and represents a loss of information from truly continuous variables). These are key advances over Mill’s methods, but QCA still requires sufficiency at the level of conjunctions of variables to reach definitive results. Also, QCA requires the inclusion of all causally relevant variables to prevent spurious inferences. In addition, the results of QCA are unstable, in that adding a single new case or changing the coding of one variable can radically change the results of the analysis.320 Moreover, because QCA assumes that various conjunctions of variables may be sufficient to an outcome, then the presence of two such conjunctions does not make the outcome any more likely or certain than one alone. In other words, in terms of the Boolean algebra upon which QCA is based, “1+1=1,” where the first two numerals are each a sufficient conjunction and the third is a positive outcome. Yet if the conjunctions in question are not fully sufficient, regardless of the values of omitted variables, a combination of two nearly sufficient conjunctions would usually be more likely to produce an outcome than either conjunction alone, barring an offsetting interaction between the conjunctions.

For these reasons, Ragin warns against the “mechanical” use of QCA for causal inference.321 He also notes that cases with the same values on independent variables may exhibit different values on dependent variables, which should spur the researcher to examine these cases more closely to determine if there are important omitted variables on which the cases differ. 322 For example, in one of the most interesting and ambitious applications of QCA, Timothy Wickham-Crowley coded twenty cases of actual and potential peasant support for guerrilla movements in Latin America, but did not focus on several cases with similar independent variables but different outcomes. Closer examination of these cases might have identified omitted variables and strengthened the conclusions of Wickham-Crowley’s QCA analysis.323

In short, with QCA, as with Mill’s methods, it is necessary to supplement case comparisons with process-tracing of cases in order to relax the restrictive and unrealistic assumptions necessary for definitive results from comparisons alone.324 It may often be preferable to use the less restrictive assumptions of typological theory as a guide to the selection of cases for study and to draw contingent generalizations from these cases in ways that are less restrictive than the method of QCA.

ATTEMPTS TO ACHIEVE CONTROLLED COMPARISON THROUGH STATISTICAL ANALYSIS AND SOME ALTERNATIVES

Since perfectly comparable cases for comparative analysis seldom exist, some analysts have sought to enlarge the number of cases under study so that statistical techniques can be used. The use of statistical techniques is widely accepted in experimental settings and in social settings where the assumption of unit homogeneity is unproblematic (in other terms, when large numbers of like cases are available). However, the use of statistical techniques on small numbers of cases is more limited and involves sharp trade-offs. To “increase the number of cases” so that statistical techniques are possible, researchers must often change the definitions of variables and the research question and must make assumptions of unit homogeneity or similarity of cases that may not be justified.

A remedy often proposed is simply to redefine and broaden the research problem to make it possible to identify a large enough number of cases to permit statistical analysis. For example, Neil Smelser has suggested that the investigator may resort to the “replication of the suspected association at a different analytical level” to multiply the number of observations at another level of analysis.325 As an illustration of this practice, Smelser cited Emile Durkheim’s study of suicide in the military.