Comparative-historical methodology.

INTRODUCTION

Recent years have seen a surge of publications concerning the methods used in comparative-historical analysis. (1) These works reflect a growing self-consciousness about research design among comparative-historical analysts, and they address a wide range of issues concerning descriptive and causal inference that are of general importance to the social sciences. Although these studies have not yet had a large impact in the field of methodology, which is oriented toward statistical analysis, (2) I argue that their insights deserve a central place within social science methodology.

This argument is developed over three sections. The first two sections consider methods of causal inference, focusing respectively on tools for analyzing necessary and sufficient causes and tools for the study of temporal processes. The third section is concerned with descriptive inference, exploring techniques of conceptual innovation and tools for achieving measurement validity. In all these discussions, the emphasis is on the distinctive contributions of comparative-historical methods-that is, contributions that go beyond what mainstream statistical methods have to offer. (3) The article closes with a call for assigning comparative-historical methodology a more central place in general social science methodology.

TOOLS FOR STUDYING NECESSARY AND SUFFICIENT CAUSATION

Hypotheses about necessary and sufficient causes--including probabilistic necessary and sufficient causes--are commonplace in nearly all domains of research. However, to evaluate such hypotheses, researchers cannot rely on mainstream statistical tools. Standard regression frameworks will incorrectly estimate causal effects when confronted with these kinds of causes (see Braumoeller & Goertz 2000, Ragin 2000). By contrast, comparative-historical methodology offers tools well adapted to the analysis of necessary and sufficient causes.

Overcoming Skepticism

Although many statistical researchers may concede that they lack sophisticated tools for identifying necessary and sufficient causes, they likely will argue that necessary and sufficient causes are not relevant to the social sciences, and thus that this deficiency is not a problem. The belief that necessary and sufficient causes are irrelevant to the social sciences is common among methodologists. It is therefore useful to discuss these causes in relationship to skeptical concerns about them.