It is our impression, however, that much rational choice theory is method driven rather than problem driven, and that this is partly responsible for its defects. Empirical science is problem driven when the elaboration of theories is designed to explain phenomena that arise in the world. Method-driven research occurs when a theory is elaborated without reference to what phenomena are to be explained, and the theorist subsequently searches for phenomena to which the theory in question can be applied.
Green, Doland and Ian Shapiro. Pathologies of Rational Choice Theory: A Critique of Applications in Political Science. 1994. p. 194 (hardcover)
Method driven research sounds like theology, with practitioners engaged in internal arguments about theories that bear no necessary relation to phenomena in the real world
“Fifth, it is argued that formal models offer too abstract an account of human behavior (Shapiro 2005). Formal models are complex exercises that derive propositions mathematically from an initial set of assumptions. These assumptions may be questionable or may be oversimplifications. The setting out of a set of relationships in this way requires many steps and proofs. With such a mode of analysis, these models can appear to be perverse, showing a liking for complexity and obscurity for its own sake, indulging in increasingly implausible claims and hypotheses, or claiming moves toward equilibrium points that do not seem to conntect to the real world of decision making but are only understood from within the model. This is what Shapiro called method-driven research, vastly inferior to problem-driven research (Shapiro 2005). The technical side and its admiration have taken over to the detriment of using social science to solve practical problems. The criticism is particularly troubling for the student of public policy who wants to be grounded and connected to the real world. The response to the criticism is to say that some formal models do seem too abstract and complex, but not all of them are. This is because there is now much more attention in formal modeling circles to tests of rational choice theory, making empirical relevance one of the key criteria for selecting models (Morton 1999). The annual workshop Empirical Implications of Theoretical Models has run since 2001 and is devoted to empirical tests of these models. Moreover, most treatments using formal models tend to contain both the model and the test (for example, Keefer and Stasavage 2003). These tests use modeling and sophisticated statistical treatments – but in ways that integrate the two. The idea is that the theory forces researchers to be explicit about their assumptions and to refrain from testing intuitive models where they run the risk that they retrofit their theory to the result alrady appearing in the model (though, of course, formal theories can be retrofitted as well).”
John, Peter. Analyzing Public Policy: Second Edition. 2012. p. 119