• Choose a category
• All
• Classic-Fiction

By Root 1431 0

strict requirements of controlled comparison (though they acknowledge its utility if carefully matched cases provide adequate control). 338 Yet they do not raise in any detail the other ways of meeting the requirements of controlled comparison that are discussed earlier in this chapter. Instead, they propose a different method for assessing theories, one that focuses almost exclusively on the observable implications of a theory for independent and dependent variables, but with little attention to intervening variables or within-case means of assessing them (with one exception—a discussion of research on the extinction of dinosaurs). DSI expresses a strong preference for this alternative, which forms the centerpiece of the book.

The method of testing theory by its observable implications for independent and dependent variables is indeed an alternative, a familiar one in discussions of methodology. Since the authors claim that this method fills a major gap in qualitative methodology, it deserves to be taken seriously and to be subjected to questioning, which follows below. Critical reviews of Designing Social Inquiry focus mainly on other issues, and we will not summarize all the questions these reviews have raised.339

Although DSI does not refer to the problem of “too many variables, too few cases” in so many words, it recognizes the importance of the problem, calling it “the Fundamental Problem of Inference.” This problem exists because “we cannot rerun history at the same time and the same place with different values of our explanatory variable each time—as a true solution to the fundamental problem of causal inference would require.”340 This statement conveys a recognition of the great difficulty of employing the experimental method for analyzing historical cases. It also is the foundation for their quite measured view of the feasibility of controlled comparison.

DSI discusses “two possible assumptions that [in principle] enable us to get around the fundamental problem.” They emphasize that these assumptions, “like any other attempt to circumvent the Fundamental Problem of Causal Inference, always involve some untestable assumptions.”341 One of these assumptions is that of “unit homogeneity—the assumption that “two units [cases] are homogenous when the expected values of the dependent variables from each unit are the same when our explanatory variable takes on a particular value.”342 At the same time, however, DSI recognizes that such an assumption is often unjustified; two cases “might differ in some unknown way that would bias our causal inference.”343 To this qualification we add that the assumption of unit homogeneity is not justified when the phenomenon in question is affected by equifinality—i.e., when similar outcomes on the dependent variable have different causes.

DSI maintains that the concept of unit homogeneity (or a somewhat less demanding assumption of “constant causal effects”) “lies at the base of all scientific research.”344 However, this assertion is not squared with the prevalence of equifinality. In stating that this assumption underlies “the method of comparative case studies,” DSI overlooks the fact that the comparative method, combined with process-tracing, can be and has been employed to analyze and account for differences between cases—that is, cases that do not exemplify unit homogeneity.

In fact, in the end DSI does not place much confidence in the validity and usefulness of the unit homogeneity assumption; the authors say that obtaining it “is often impossible” and it is important for researchers to understand the “degree of heterogeneity” in units examined and estimate as best they can “the degree of uncertainty—or likely biases” that must be attributed to any inference drawn from the comparison.345 Such a statement overlooks once again that when equifinality is present, a different procedure is necessary instead of an effort to assess the degree of uncertainty involved in comparing the cases.

It is clear, however, that DSI shares our belief that the requirements for strict, controlled