King, Keohane, and Verba's Designing Social Inquiry (1994) is a seminal book outlining how scholars might transfer of the logic of quantitative causal inference to qualitative research. It is an important guide for both teaching and research, providing us with an initial set of tools for constructing research designs that yield maximum explanatory power.

Such easy-to-remember maxims as "don't select on the dependent variable," "sample on the independent variable," and "the more observable implications of your theory you can test, the better," are readily digestible and easy to teach, and for that reason have quickly become part of the standard vocabulary of most comparativists.

As useful as these tools (and others that KKV suggest) are, we maintain that they constitute only a small part of what makes for a good research design in comparative politics. Indeed, when we examine the classics in comparative politics, we find that the relationship between theory and evidence is not what KKV would lead us to expect.

In this essay we will argue that comparativists today should begin to combine the rigorous methodological standards set out by KKV with the analytic focus on regime types and institutional diffusion characteristic of earlier classics in the field. First, we will show how insufficient attention to the evolution of regimes over time and space tends to produce two interlinked methodological problems: conceptual stretching and "Galton's problem."