This major new book originated in a seminar on American politics that Gregory Wawro and Ira Katznelson taught together at Columbia University. The two scholars come from different parts of the discipline: Wawro is a quantitative political scientist trained in contemporary statistical approaches to causal identification and inference, whereas Katznelson is a qualitative and historical researcher associated with American political development. Yet they share the common goal of generating social science work that adopts both the rigorous generalizing ambitions of quantitative analysis and the focus on case particularity, context, and temporality that is characteristic of historical analysis. The resulting manuscript is the latest installment in what is now a series of works in which scholars (predominantly men) associated with the qualitative and quantitative traditions join forces and offer methodological commentaries and advice (the first and still most famous of these is KKV, i.e., King, Keohane, and Verba’s Designing Social Inquiry, 1994).

Within this stream of research, Wawro and Katznelson’s contribution stands out for its focus on how new quantitative methods can be used to study politics in contextually and temporally sophisticated ways. Following in the footsteps of the sociologists Larry Isaac and Larry Griffin, Wawro and Katznelson argue that mainstream quantitative methods can and should be redesigned to overcome their ahistorical and atheoretical tendencies. They dispute the notion that quantitative research is necessarily decontextualized, and they move forward with the positive agenda of developing concrete tools that quantitative scholars can use to study context and temporality in ways consistent with historical research.

Wawro and Katznelson see the need for historical and contextual work as rooted in the fundamental fact that causal heterogeneity is ubiquitous in the social world. The authors insist that homogeneity assumptions are routinely not sustainable; they argue that social scientists must embrace this reality rather than downplay it. Wawro and Katznelson go so far as to argue that, “unit heterogeneity is a desirable thing and should be made a constitutive part of historical analysis. Instead of perceiving it as limiting an analysis, embracing unit heterogeneity and adjusting models to accommodate it is essential for the empirical component of historicizing theory” (p. 15). This declaration alone is likely to win over many qualitative methodologists (including this reviewer). Moreover, by starting from the assumption of causal heterogeneity, Wawro and Katznelson clearly distinguish themselves from KKV and the effort to use additive-linear regression as a tool for thinking about and designing research.

The authors begin with a sober recognition of the limitations of most existing quantitative approaches. With respect to additive-linear models, they argue that their functional form is simply not an accurate depiction or often even approximation of historical reality. Although they appreciate the efforts of scholars to use “moving regression” techniques and the like, they believe that these tools are still inadequate and reject the frequentist logic on which they are built. Wawro and Katznelson instead use Bayesian approaches to enlarge the standard statistical toolkit. Their advocacy of Bayesianism is practical: They believe that Bayesian assumptions offer a more useful foundation from which to bring essential qualitative insights into quantitative analysis.

Wawro and Katznelson discuss several quantitative techniques that incorporate time-variant features of relationships and that capture differing degrees of parameter heterogeneity. They favor nonparametric regression approaches that assume complexity and inconsistency in covariates and that use function smoothers to summarize local patterns in the data. Given that some variables may operate in the standard additive linear way, the authors call for the Bayesian estimation of semiparametric regression (rather than only nonparametric regression). While these tools may be familiar to quantitative methodologists, the authors’ key point is that they are only rarely used in practice. Most quantitative political scientists use methods that fail to incorporate temporality in sophisticated ways. Time Counts has as its mission the advancement of scholarship by inspiring both greater awareness and actual use of these techniques.

The book includes a chapter on path dependence, focusing on self-reinforcing sequences in which initially stochastic outcomes become locked in. To model path dependence, Wawro and Katznelson explore parameter variation techniques and change point models that can represent shifts in relationships that are abrupt and discontinuous rather than gradual and smooth. Just as qualitative researchers study patterns of both gradual and sudden institutional change in the political world, so too can quantitative researchers identify both rapid and gradual patterns of change in causal relationships within data sets. The authors devote special attention to Markov switching models as useful tools for assessing aspects of theories of path dependence. They do so both abstractly and by carrying out empirical replications of existing work on American politics.

Different kinds of scholars are likely to react to this book in different ways. Scholars who view the qualitative and quantitative research traditions as separate cultures defined by distinct research goals and contrasting research questions are likely to see the book as a good faith effort to advance the quantitative research tradition. On this view, which I hold, the book is a successful discussion of ways to improve research that asks questions about causal effects for large numbers of cases. However, I believe the book does less to advance methods in the qualitative tradition, where I see the goal as the study of actual causality and the development of theories of the causes of specific outcomes in particular cases.

A more radical criticism is likely to come from interpretive scholars, including many historians, who are often interpretive in their goals. Here we must recognize that, from the standpoint of history, the discipline of political science can be seen as obsessed with a narrow understanding of causality derived from statistics. I suspect that many historians will see this book as undervaluing the importance of noncausal interpretation based on deep contextual knowledge, normative sensibilities, and narrative explanation. Because most historians do not pursue the goal of estimating the causal effects of variables, the authors’ proposed solutions may not seem relevant to their own work.

Finally, scholars who advocate for mixed-method research may want to build on this argument by considering the ways in which qualitative research can enhance the statistical techniques that the authors propose. On the mixed-method view, quantitative research can almost always be improved with qualitative research, just as qualitative research can almost always be improved with quantitative research. Scholars in favor of mixed-method research may support the goals of Wawro and Katznelson to enhance causal inference, but they will probably insist that the authors do not say enough about the ways in which qualitative insights are needed to estimate the proposed semiparametric regression models under Bayesian assumptions. They will press for an integrative program in which scholars go back and forth between quantitative techniques and qualitative data.

Although overall reactions will vary, I think all political scientists stand to learn from this balanced and nonpolemical effort to reorient quantitative analysis for historical social science. One of lessons to be learned is found precisely in the deeply respectful and positive approach that the authors bring to what is too often viewed as an antagonistic disputation between qualitative and quantitative research.