Commentary: The Strategic Value of Analytic Narratives

Since its publication, the Analytic Narratives volume has stirred considerable interest. In less than three years, it has received both admiring approval and blistering criticism, and both categories have ranged from judgment passed on the general enterprise to detailed critiques of the individual chapters. As someone who is sympathetic to the approach and the issues explored in the specific cases, I will take a slightly different tack. Beginning from the premise that the Analytic Narratives method has a unique and important contribution to make to historical social science research, I will explore how the authors might improve the argument’s layout so that others can more effectively adopt their methods. In other words, how do we go from an impressive collective of exemplary pieces to a research strategy whose characteristics are articulated clearly enough to be used by scholars with analogous research questions?

The AN approach has two components: the “analytic,” which in this context means the use of rational choice tools, and the “narrative,” which here means a specific qualitative historical presentation of the case data for the question to be analyzed.¹ The authors emphasize the advantages of the rational choice framework, but they shy away from stating what makes rational choice best suited for the questions that interest them. They instead suggest that “a range of models could serve as the basis for analytic narratives” (3), including those derived from new institutionalism or analytic Marxism. They also posit that analytic narratives complement structural and macrolevel analyses rather than superseding them (13). These statements are overly modest; they understate the distinctive contribution of their method. Other approaches clearly have their own advantages and drawbacks. But if the authors do not make a convincing argument for the greater effectiveness of analytic narrative in explaining a given problem, it loses some of its power. In important ways, analytic narratives and macrostructural analyses are not complementary but competing. As the authors themselves point out, analytic narratives emphasize agency and choice, while macrostructural approaches emphasize historical conjunctures, social and political cleavages, and the role of institutions in shaping actors’ preferences over outcomes. It would be more helpful and more accurate to discuss the conditions under which each is more effective. For example, analytic narratives’ assumption of actors making strategic choices may more accurately explain outcomes in particular contexts or time periods, while macrostructural approaches are more suited to explaining the shift from one set of societal equilibria to another.

A similar instance of underspecification occurs in the discussion of the use of narrative as the best way to organize and present the empirical material. There exists a growing body of research that combines rational-choice-based formal theoretical models with a variety of data, including qualitative non-narrative and quantitative data. These empirical approaches also address the Donald P. Green and Ian Shapiro (1994) criticisms concerning the supposed lack of empirical tests of formal models. The reasons behind the choice of a narrative case method to pair with rational choice models need to be made more explicit, beyond the authors’ observation that since Thucydides, “the narrative form has been the dominant form for explaining human behavior” (12). Over the last century, the use of qualitative analyses of structural conditions in which behavior takes place, and especially the development of systematic quantitative measures, has at times supplanted the narrative form in political science, economics, psychology, and even sociology. Given these trends, what makes an emphasis on narrative justified as an analytically powerful approach? What is its distinctive contribution? In a short section in the introduction, the authors point to at least one important reason: the questions to be analyzed appear to be dependent on dynamic processes—that is, the outcomes of interests developed over time. Without explicitly invoking time-sensitive concepts such as path dependence, the authors all focus on cases that are characterized, if not defined, by a temporal dimension. In each case, actors’ choices are shaped by decisions made in...