This edited volume makes four different arguments, which to this reviewer varied in their persuasiveness. In what follows I consider these positions in turn.

The central argument of Interpretive Quantification is a straightforward claim about methodology. Editors J. Samuel Barkin and Laura Sjoberg define methodology as “the analysis of what methods are chosen and why, given certain assumptions” (p. 4). They distinguish methodology from methods, which they define in terms of specific techniques of information gathering and analysis. The volume makes the methodological claim that, contrary to usual assumptions and practice, quantitative methods are often appropriate for constructivist and critical-theory work in international relations.

In this respect, the volume is aligned with Gary King, Robert Keohane, and Sidney Verba’s Designing Social Inquiry (1995), which Barkin and Sjoberg refer to as “KKV.” The two works are aligned in that they both explore the relationship among method, methodology, and research design, and both seek to show that different methods are compatible with work they regard as important. KKV argues that qualitative social research can be conducted along the same principles of inference as quantitative research. Interpretive Quantification seeks to show that quantification is compatible with critical and constructivist social analysis. As Barkin and Sjoberg state, “this project is at least partly about taking seriously KKV’s observation about the logic of inference transcending the methods used” (p. 245).

Most of the volume is devoted to substantiating this central argument. Brooke Ackerly, Cameron G. Thies, Matthew Hoffmann, and Paul A. Kowert illustrate how statistical methods and network analysis can be used in critical or constructivist studies of human rights, conflict studies, emissions trading, and images of various countries. David Sylvan, Ian S. Lustick, Barkin, and Sjoberg with Kevin Knudson explore the value of formal and computational methods for understanding diplomatic conversations, forecasting the future, governing international fisheries, and interpreting democracy. In his chapter toward the end of the volume, Patrick Thaddeus Jackson offers his own interpretation of the relationships between what he calls theory, methodology, and method. His chapter and the editors’ conclusion comment, in some ways critically, on the other contributions to this volume. Together, these diverse and interesting essays strongly support the editors’ claim that critical and constructivist studies can benefit in certain situations from quantification, and therefore that quantification methods are compatible with constructivist or critical theory and methodology. In its principal purpose, therefore, Interpretive Quantification succeeds.

In his contribution, Jackson goes further, arguing that theory, methodology, and methods can be combined in a variety of ways, well beyond the coupling of critical and constructivist theory and quantitative methods highlighted in this volume: “The potential number of other such combinations is only limited by the number of theories, methodologies, and methods available to be combined so even if we limited ourselves to three families of theory (the IR ‘isms’), four methodologies (neopositivism, critical realism, analyticism, and reflexivity), and three families of method (numerical, textual, and participant-observation), we would still be talking about thirty-six different combinations of theory, methodology, and method” (p. 232).This is an imaginative and creative observation, which I hope will stimulate productive research employing various such combinations.

The second argument of the volume is that what the editors label as the “neopositivist” analysis in KKV is a “specific narrow epistemology of social science” (p. 9), rather than a general set of methodological rules applicable to a variety of methods and usable by social scientists with different theoretical orientations. Whereas KKV seeks to unify political science inquiry, they argue for its “disunity” (p. 11). Indeed, they seem to believe that constructivist analysis, as well as the enterprise of critical theory, can only be carried out if liberated from the straightjacket of KKV’s narrow epistemology.

The editors' argument for the disunity of social science begins with the proposition that the social sciences are recursive: “[T]heory not only organizes observations about the object of study but actually changes that object” (p. 8). They claim that KKV’s “phenomenalist” epistemology, relying on observable facts, is therefore incompatible with understanding international politics.

It seems to me, on the contrary, that constructivist theory is entirely compatible with the discussion of inference in KKV. Constructivist theory assumes that ideas and intersubjectivity play crucial roles in constructing what the social world is (p. 17). As I argued in my 2000 Presidential Address to the American Political Science Association, “both money and prestige matter a great deal in politics, but only insofar as people hold beliefs about others’ beliefs” (“Governance in a Partially Globalized World,” American Political Science Review, 95(1), 2001: 8). In this way, social science is different from natural science, and nothing in KKV implies otherwise. International politics is socially constructed, and for this reason existing practices are potentially reversible, as those of us who study American foreign policy and multilateral institutions now observe, to our chagrin.

These observations, however, are not conclusive about our epistemological and methodological choices. The potential recursiveness of social science does not pose an obstacle to social scientists seeking as objectively as possible to make descriptive and causal inferences about what they observe. Of course, social reality changes, and so all of our inferences are conditional on conditions existing in the time period under observation. These inferences are also conditional on many other factors, however: Social science is much more about conditional generalizations than general laws.

If change in world politics were both rapid and random, we would have to give up the social science–IR enterprise, since what we observed today would be irrelevant tomorrow. But while change may be more or less rapid, behavior is not random and patterns of behavior tend, in somewhat different forms, to recur. We can therefore make inferences from the past—whether about interstate conflict, coalitional alignments, or patterns of cooperation—that can provide valuable insights—although they rarely produce reliable predictions—in the present. Constructivists, therefore, can generate hypotheses that are compatible with their theoretical perspectives and test them by examining their observable implications, just as realists and institutionalists can.

The third major argument is that constructivism and critical theory can be properly grouped together as alternatives to neopositivism. Yet when Barkin and Sjoberg discuss critical theory, they generate confusion and engage in “concept stretching” (Giovanni Sartori, “Concept Misformation in Comparative Politics.” American Political Science Review, 64(4), 1990: 1033–53).They describe critical theory not as a “tool”—used to help us understand the world—but as “critique” and “practice.” But if theory is now defined as critique and practice rather than a tool to help us understand the world, the meaning of theory has changed. Critique and practice are fundamentally different enterprises than scientific understanding. Coupling constructivism and critical theory therefore seems to me to result in a category mistake. Constructivism has much more in common with realism and institutionalism than with critical theory. Valuable as it often is, critical theory is more a normative and activist enterprise than descriptive or explanatory social science.

The fourth major argument concerns the role of facts in social science. Jackson criticizes KKV for taking facts as given, focusing on the inferences to be made from them. He is surely correct that neither qualitative nor quantitative hypothesis testing is necessarily more scientific than “the careful ethnographic description of a form of life undertaken in order to make plain the hidden or tacit social rules operating in that particular situation or community” (p. 238). “Making plain the hidden or tacit social rules,” however, clearly involves descriptive inference, not simply description, since social rules are enduring and not directly observable. Jackson seems to imply that those who insist that social science requires explicit and transparent rules of inference are acting as “gatekeepers” and “disciplinary police,” but his own argument implicitly invokes inference and therefore implies a need to explore which modes of inference are valid.

Barkin and Sjoberg criticize KKV for taking empirical observations “unproblematically as ‘facts’” and seem to take seriously the notion of a “transfactual” social science (p. 7). Yet no good social scientist assumes that any empirical observation is unproblematic: We are all aware of observational biases and errors. Designing Social Inquiry distinguishes descriptive inference from discovery of facts; but we were not so naive as to believe that facts themselves are unproblematic. In his own chapter, Barkin claims that KKV view social science merely as hypothesis testing, although it is hard for me to understand this interpretation in view of my own work on multilateral institutions. Like Barkin, I use rational choice theory as a deliberate oversimplification in order to illuminate modes of human action that do not fit the rationalist model (see Robert O. Keohane, After Hegemony: Cooperation and Discord in the World Political Economy [1984], Chapter 7, especially pp. 108–9, and Chapter 7, pp. 111–32, which discusses bounded rationality, changes in preferences, and empathy).

What is “narrowness” from one perspective is careful delineation of the boundaries of social science from another. I am not bothered by the criticism that KKV is too fact-bound. On what can social science rely if not on facts, and the inferences that we can make from them? The president of the United States in 2017 has repeatedly made incorrect statements that his advisors characterize as “alternative facts,” but social scientists need more solid grounds for their statements, so that their facts can be checked by their peers. In other words, facts are fundamental to social science, and the notion of a “transfactual” approach makes even less sense in political science than in politics.

Interpretive Quantification is a carefully argued and thought-provoking volume. It directly attacks the conventional dichotomy between positivist quantification, on the one hand, and nonpositivist qualitative analysis, on the other, and should therefore help call scholars who continue to rely on this dichotomy to account. Its arguments for the disunity of science, for linking critical theory with constructivism, and for a “transfactual” social science seemed to me much less persuasive. Yet even those of us not convinced by these arguments should be grateful to Barkin and Sjoberg and their collaborators for making us think.