Case Studies and Theory Development in the Social Sciences - Alexander L. George [70]
Still, the reflexivity of social subjects does constrain social science theorizing in a variety of ways. Strategic interaction, self-fulfilling and self-denying prophecies, moral hazard, selection effects, and a range of other phenomena make the development of predictive theories far more difficult in the social sciences than in the physical sciences.253 To cope with these difficulties, social scientists should distinguish here between theories that can explain and predict both processes and outcomes, which are common in the physical sciences, and those that can explain processes and outcomes but not predict them. The second kinds of theories, common in the social sciences, are also found in the physical sciences. Theories of evolutionary biology, for example, explain processes and post facto outcomes, but they do not predict outcomes. While social scientists should aspire toward predictive theories—our own approach to the development of typological theories is meant to foster contingent generalizations with predictive (or at least diagnostic) power—they should also recognize the value of good historical explanations of cases as well as that of law-like generalizations.254 The two are linked, in that historical explanations use theoretical generalizations to argue why in a particular context certain outcomes were to be expected, and good historical explanations (especially of cases with surprising outcomes) can lead to the development of better theories.255 But the logic of historical explanation does differ from that of nomological generalization, as we note in Chapter 10.
These factors give theories in the social sciences a different “life-cycle” from those in the physical sciences. In the social sciences, much cumulation takes the form of increasingly narrow and more contingent (but also more valid) generalizations.256 At the same time, fundamental changes in research programs are more frequent in the social sciences than in the physical sciences not just because of the “faddishness” or “subjectivity” of the social sciences, but because the objects of study change in reflexive ways.
In sum, while the reflexivity of human agents means that the philosophies of the social and physical sciences differ in important ways, we argue that progressive theorizing over long periods of time is possible in the social sciences. We also urge more explicit differentiation of various types of theory. We concur, moreover, with the scientific realist view that social facts exist independently of the observer and can be the subject of defensible causal inferences.
Theoretical Explanation: From the Deductive-Nomological Model to Causal Mechanisms
The traditional positivist model of explanation associated with Karl Popper, Karl Hempel, and Ernest Nagel posits that a “law or event is explained when it is shown to be something that is or was to be expected in the circumstances where it is found.”257 In this view, laws, or covering laws, are statements of regularity in the form of “if A, then B,” and explanation consists of combining a law with initial conditions A to show that B was to be expected. While this model of explanation, developed by Hempel and Paul Oppenheim and later labeled as the “deductive nomological” or “D-N” model, remains intuitively appealing and widely used, it suffers from several serious shortcomings. First, it does not distinguish between causal and spurious regularities. Second, it does not indicate whether the outcome B was to be expected with 100 percent certainty or something less than certainty.258 In this section, we elaborate briefly upon each of these challenges and indicate how they have driven philosophical debate toward the scientific realist notion