• Choose a category
• All
• Classic-Fiction

By Root 791 0

methods detailed in subsequent chapters, particularly case comparisons and process-tracing. We conclude, therefore, by emphasizing three practical implications for qualitative methods. First, regarding the goals of social science research, we do not need to view ourselves as the poor relatives of the physical sciences or eschew attempts at causal and potentially even predictive theories. The changing and reflective nature of social subjects makes social science theories more provisional and time-bound than those in the physical sciences, but does not prevent cumulative and progressive theorizing over long periods of time in the form of middle-range theories. Theoretical progress in the social sciences can consist of advances in puzzle-driven research programs, increasingly complete and convincing historical explanations, and theories that are stronger at explaining social behavior than at predicting it. Progress is not limited to the development of general theories or schools of thought with greater validity, scope, or predictive capability—as desirable as these kinds of progress may be.

Second, an attempt to explain by reference to causal mechanisms, which in principle requires consistency with the finest level of detail observable, provides a powerful source of causal inference when carried out through the method of process-tracing, which examines processes within single cases in considerable detail. In practice, process-tracing need not always go down to the finest level of detail observable, but by avoiding as if assumptions at high levels of analysis and insisting on explanations that are consistent with the finest level of detail observable, process-tracing can eliminate some alternative explanations for a case and increase our confidence in others.

Finally, typological theorizing, which combines methods of both cross-case comparison and process-tracing, is a powerful way to create middle-range theories that are consistent with both the historical explanations of individual cases and the general theoretical patterns evident across cases. Such theorizing makes very limited assumptions about whether a causal mechanism operates in similar ways across different contexts. At the same time, typological theorizing attempts to outline the conditions under which a particular causal mechanism has a defined effect, and the differing effects it has in different contexts, by modeling recurrent combinations and interactions of mechanisms. In short, typological theorizing offers the promise of cumulation without losing sensitivity to context.

Chapter 8

Comparative Methods: Controlled Comparison and Within-Case Analysis

In this chapter we discuss comparative methods—the case study methods that attempt to approximate the conditions of scientific experiments. Comparative methods involve the nonstatistical comparative analysis of a small number of cases. Perhaps the best known and still dominant variant of comparative methods is controlled comparison, the study of two or more instances of a well-specified phenomenon that resemble each other in every respect but one.295 When two such cases can be found, controlled comparison provides the functional equivalent of an experiment that enables the investigator to make use of experimental logic to draw causal inferences. This possibility gives controlled comparison considerable appeal.

Yet such control is very difficult to achieve.296 Researchers urgently need an alternative to the experimental paradigm for several reasons. It is generally extremely difficult to find two cases that resemble each other in every respect but one, as controlled comparison requires. The familiar alternative of using statistical analysis to achieve the functional equivalent of an experiment often runs into the problem that there is an insufficient number of cases for many phenomena of interest.297 Even in the relatively few cases in which genuine experimentation is possible, it is often ethically problematic and sometimes forbidden.

Many writers have noted that the impossibility of applying experimental