In this groundbreaking book, Orit Kedar develops an original and important argument that has fundamental implications for how political scientists analyze voting behavior and election outcomes. The standard Downsian perspective is that voters support parties whose policy programs mirror the voters' own policy preferences so that center-left voters support center-left parties, right-wing voters support right-wing parties, and so on. In the introductory chapter of her book, Kedar notes that this proximity-based model cannot explain the following empirical patterns: that voters in parliamentary democracies frequently support parties whose policy positions are more radical than those of the voters who support them; that in presidential democracies, the president's party typically fares worse in nonconcurrent elections than in elections held concurrently with the presidential contest; and, that in federal systems such as Germany, governing parties typically fare poorly in subnational elections.

To illuminate these patterns, Kedar presents theoretical and empirical analyses in support of a provocative thesis: namely, that voters decide in part on the basis of their projections of how the competing parties influence policy outputs—a thesis she labels the compensatory vote model—instead of evaluating parties entirely on the basis of representational considerations relating to parties' policy promises. She argues that in political systems such as Britain, where policymaking power is typically concentrated in the hands of a single party with few checks and balances, both representational and compensatory voting considerations should prompt voters to endorse parties whose policies mirror the voter's own viewpoints. However, in systems such as the Netherlands that feature extensive power-sharing among parliamentary parties, and also in presidential systems such as the United States that often feature power-sharing between representatives of different parties in the executive and the legislature, compensatory voting considerations may motivate voters to “overshoot” on policy by supporting parliamentary parties whose positions are more extreme than the voter's own policy viewpoints, since these noncentrist parties may do the most to pull policy outputs in the voter's preferred direction.

Chapter 2 presents a voting model that combines these representational and compensatory vote perspectives, and develops applications to legislative elections in parliamentary, presidential, and federal systems. The key feature of the compensatory component of the model is that voters engage in a counterfactual thought experiment whereby they evaluate the party's policy impact by comparing policy outputs when the party is in parliament versus expected policy outputs with the party entirely absent from the policymaking process, and that voters give their highest policy-based evaluation to the party that has done the most to move policy outputs close to the voter's ideal point. Thus, the compensatory model specifies that citizens' party support reflects their assessments of the party's overall policy impact, that is, the difference between policy outputs with the party in parliament versus outputs with the party entirely absent from parliament, rather than the instrumental assessment of the marginal impact of an individual citizen's vote via its effect on the composition of the parliament (where this vote can be expected to add—at most—one additional seat to the preferred party's seat share). Kedar embeds this compensatory vote model into a more general model in which voters have a mixture of compensatory motivations along with representational (i.e., expressive) motivations to support parties that advocate policies in line with the voter's own beliefs.

Chapters 3–5 report empirical analyses of voting in a variety of institutional contexts, which evaluate the extent to which voters weigh representational versus compensatory considerations. Chapter 3 reports individual-level analyses of election survey data from 14 Western parliamentary democracies that support the hypotheses that voters weigh both types of considerations and that, crucially, the more compromise and power-sharing facilitated by parliamentary norms and procedures and by the electoral system, the more voters engage in compensational voting to account for the fact than any single party's policy influence will be diluted by the other parliamentary parties. Chapter 4 reports aggregate-level analyses of election outcomes in 34 presidential democracies, which demonstrate that—consistent with the compensational voting model—the party of the president suffers greater vote losses in nonconcurrent elections (when voters can more easily balance the policy orientation of the legislature against those of the president) than when the legislative election is held concurrently with the presidential election, and that the greater the policymaking powers of the president relative to the legislature, the greater the loss suffered by that party in midterm elections. Chapter 5 reports aggregate-level analyses of the outcomes of subnational (Lander) elections in Germany as a case study of voting behavior in federal systems. Chapter 6 concludes with a discussion of compensatory voting across different institutional regimes and implications for party strategies and for political representation.

I am extremely impressed with this book on both theoretical and empirical grounds. Theoretically, the author's development of the compensatory vote model—and its integration with the standard, proximity-based, expressive voting model—represents an important extension of earlier policy-based voting research including George Rabinowitz and Stuart Elaine Macdonald's directional model, Torben Iverson's representation policy leadership model, and (most directly) Bernard Grofman's policy discounting model, along with American-based research on divided government by Morris Fiorina, Alberto Alesina, and Howard Rosenthal, and others. Furthermore, this book is comparative in the best sense of the word. Kedar systematically works out nuanced implications of the compensatory voting model for alternative institutional contexts, and in empirically evaluating these arguments, she examines elections and voting behavior in an extremely wide range of political systems; as discussed, her individual-level analyses of compensatory voting in parliamentary democracies encompasses 14 Western democracies, while her aggregate-level analyses of policy balancing in legislative elections in presidential democracies encompasses 199 elections in 34 presidential democracies.

The author is also meticulous in discussing and evaluating alternative explanations for the empirical patterns she uncovers, and at numerous points she describes additional behavioral implications of her model that can be tested in future research. In so doing, she demonstrates a magisterial grasp of the institutional workings of the world's democracies and a fine-grained sense of how alternative institutional contexts affect citizens' compensatory voting incentives, which she marries to wide-ranging and creative empirical tests of hypotheses derived from the compensatory voting model.

Overall, Kedar makes a persuasive case for the compensatory voting model. In my opinion, what is convincing about her findings is not any single statistical test (since, as she carefully notes, many of the statistical patterns uncovered are compatible with alternative explanations, some of which cannot be evaluated) but the cumulative weight of her analyses. For while any single empirical test may be problematic, the fact that the wide variety of analyses reported in this manuscript—which range over different levels of analysis, different countries, and different institutional contexts—all point in the same direction weighs heavily in favor of the compensatory vote model. Voting for Policy, Not Parties thereby makes immense contributions to the literatures on voting behavior and election outcomes, and, as Kedar notes in the concluding chapter, the compensatory voting model also has important implications for parties' policy strategies and for political representation. This compelling, wide-ranging book will influence research on these topics for years to come.