Adams's main point of departure is that while voters in advanced democracies maintain partisan loyalties over decades, many citizens, and not only independents, vote on the basis of the issue positions of candidates and parties. Note that these two facts are not contradictory, as he discusses at length. Furthermore, spatial voting models have not been used often or successfully to understand multiparty systems where voters have these enduring partisan loyalties.

Adams begins with a fairly standard spatial voting model. Parties compete over a single, ideological dimension, they can change positions to attract voters, and voters choose parties weighing both their partisan biases and the ideological distance between their ideal points and the positions of the parties. When he combines these basic assumptions with other features of party competition, he enters challenging theoretical territory. The set of assumptions used for any given chapter corresponds to the case at hand. For instance, he analyzes a spatial model with four parties competing in his first of two chapters on French elections, whereas he analyzes a spatial model with three parties competing in his chapter on British elections.

After describing the basic model, Adams repeats a pattern several times in the book, first for a generic case, then for Great Britain, then for France, and finally for the United States. The pattern proceeds as follows: He uses survey data to show a correlation between issue preferences and partisanship; he proposes a hypothetical distribution of voters' preferences and partisanship; if possible, he derives analytical results on the behavior of the parties given the hypothetical distribution of preferences; he simulates party strategy trajectories using computational models, again relying on the hypothetical distribution of preferences; then he simulates party stategy trajectories using a distribution derived from real data on voters' preferences; and finally, he compares the party strategy trajectories derived from the simulations to trajectories drawn from real data on party strategies over the last several decades.

One can only be impressed by two aspects of this research project. First, the project has a creative mixture of methodologies and data, bringing to bear mathematical models, computational models with simulated data and then calibrated with real data from surveys of mass publics, and data on party manifestos and expert assessments of partisan ideological positions. And second, the results from the analysis are just downright compelling. The dynamic trajectories of the parties within the computational models, and even the point predictions from some of the models on actual party positions, bear striking resemblances to the real data on party behavior summarized in the book and familiar to scholars of party politics. His models capture features of party politics that few other models from past research programs have been able to capture, such as dynamic fluctuations within ideological ranges, parties chasing each other in ideological space but at a distance, and few instances of convergence of parties on the same ideological position.

In considering the set of assumptions Adams adopts, I like the trade-off he made and his approach: Because he uses computational models, he is not constrained by analytical tractability to make heroic simplifying assumptions and can analyze models with combinations of realistic assumptions. The assumptions are well justified by empirical patterns that are familiar to scholars of elections and parties. He could have taken more systematically the next step, demonstrating the robustness of the conclusions drawn from the simulations.

I have some other concerns with the book. The author wishes to separate conceptually a voter's partisanship and his or her ideological congruence with a party. That is fine in a hypothetical situation he analyzes, yet when he simulates party behavior using real data in the calibrated computational models, he incorporates partisanship and the ideological preferences of the voters drawn from surveys, without “cleaning’ the issue congruence from the partisanship. Certainly in data from real surveys, the two concepts are intertwined in complex ways, and it would require the appropriate statistical procedure to incorporate estimations of the basic (uncontaminated) partisanship of the electorates of these countries.

Also, Adams implies throughout that his project is original because it combines spatial modeling with insights derived from behavioral research. But he overstates the case. Actually, the blending of spatial modeling and the findings from behavioral research has been going on for a long time, in both directions—so political surveys regularly include spatial scales for respondents on which to place themselves and candidates, and spatial models often include valence dimensions or biases voters have toward candidates separate from spatial distances—and with many valuable results. It is true that his research is original, just not in being among the few to combine the two approaches to studying elections and campaigns. Finally, he uses such concepts as equilibrium among parties, stasis in party positions, and stability of party competition without formally defining and contrasting these terms.

Despite these shortcomings, I encourage scholars of parties and elections to pay attention to Party Competition and Responsible Party Government. Not only does Adams use some unconventional methods; he also combines methods in new ways and even pushes some of those methodologies further, such as in his simulating party strategies using voters' preferences derived from surveys. His findings and his overall conclusions will be hard for persons in any research tradition to refute. The modeling world he creates looks a lot like the political world we live in, and so the challenge is now for those who disagree with his approach to come as close as he does in predicting the broad behavioral patterns of party competition in France, Great Britain, and the United States.