Process-Tracing Methods: Foundations and Guidelines

Chapter 8. Case Selection and Nesting Process-Tracing Studies in Mixed-Method Designs

144 Chapter 8 Case Selection and Nesting Process-Tracing Studies in Mixed-Method Designs Explaining-outcome process-tracing studies are almost by definition standalone single-case studies. In contrast, single-case studies using either theorybuilding or theory-testing variants are not intended to stand alone; instead, they seek to contribute with their specific comparative advantages to what we know about a broader social phenomenon. A case study of two democracies that did not go to war despite severe conflicts of interest updates our degree of confidence in the existence of a more general causal relationship between mutual democracy and peace, at the same time shedding light on the mechanisms that explain how mutual democracy produces peace. Yet how can single-case studies contribute to our broader knowledge of causal relationships? This chapter illustrates how the theory-building and -testing variants of process-tracing can be nested into a broader research program, something that is possible given that they share the theory-centric ambition of studying systematic elements of causal relationships with other social science research methods. But nesting is possible only through case selection techniques building on cross-case inferences made using other methods. Furthermore, explaining-outcome process-tracing studies cannot be embedded in a nested analysis because of the inclusion of nonsystematic, case-specific mechanisms in explanations. As introduced in chapter 2, choosing which variant of process-tracing to employ depends on the purposes of the study. First, is the purpose of the study to explain a particularly interesting outcome, or does the study have theory-centric ambitions? If the former is the answer, then the explainingoutcome design is chosen, with the focus of crafting a minimally sufficient Case Selection and Nesting Process-Tracing Studies 145 explanation of the particular outcome. In contrast, ambitions to understand causal relationships across a population of cases lead to the choice of either theory-building or theory-testing designs. Second, when choosing between theory-building and -testing variants, are there well-developed theoretical conjectures for empirical correlations between known independent and dependent variables or conditions, or are we in a situation where there is either no well-developed theory explaining a phenomenon or where existing theories have been disconfirmed in prior empirical analysis? In the first situation, we choose theory-testing designs, whereas theory-building designs are chosen when we lack plausible theoretical mechanisms to account for outcomes. Case selection strategies in process-tracing differ for each of the three variants of process-tracing, a fact that the existing methodological literature on case selection in process-tracing has overlooked by treating process-tracing as a single method. We discuss why existing methodological prescriptions for selection strategies are not always applicable for the different variants of process-tracing (e.g., Gerring 2007a; King, Keohane, and Verba 1994; Lieberman 2005). Prescriptions are developed for case choice in the theorybuilding and -testing variants of process-tracing, in particular focusing on how these prescriptions differ from existing guidelines. In contrast, cases are chosen in explaining-outcome designs not on the basis of research design strategies but instead because the cases are substantively important (e.g., the French Revolution), although explaining-outcome studies also have the potential to contribute insights to ongoing theoretical debates in the field. Common to all three variants, however, is the need for the analyst to be explicit about the case selection strategy, detailing the reasoning behind the selection of specific cases. Given that the perfect case for a particular research purpose usually does not exist, arguments need to be put forward that substantiate why the case can fulfill the research’s goals. After discussing case selection strategies and how they enable nesting of only theory-centric process-tracing studies into broader mixed-method designs, we discuss the challenges in making the results of process-tracing research communicate with those from other research methods. Whereas methods such as frequentist, large-n statistical methods or small-n comparative methods analyze patterns of regularity between X and Y, process-tracing methods look at both X and the mechanism linking it with Y. Therefore, there is the risk that the two talk past each other, with analyses of the causal effects of X on Y potentially incompatible with analyses of the causal relationship of X and a mechanism with Y. We now turn to a discussion of case selection strategies in the three process-tracing variants. 146 Process-Tracing Methods 8.1.Theory-Testing Process...