Are the critical players in party politics best viewed as ambitious politicians or intense policy demanders? Marty Cohen and his colleagues answer this question by elaborating on two approaches to defining American political parties and explaining their major features. The conventional approach, attributed to John Aldrich and others, is to consider parties to be teams of politicians seeking or serving in office. Their alternative approach, which they associate primarily with E. E. Schattscheider, is to define parties as coalitions of groups. Many groups in society are intense policy demanders and so are motivated to form coalitions to win elections and control government. The two approaches, the authors argue, yield different accounts of the identity of the political players who create parties, determine the policies to be advocated, choose candidates and run campaigns, and make the trade-offs between the demands of activists and the larger electorate.

The authors contest the dominant argument about parties and presidential candidates of recent years—that parties have lost control over nominations as campaigns have become candidate-centered and candidate-run affairs. Instead, they argue that the group coalitions that comprise parties have adapted to the new rules governing presidential nominations and continue to dominate the process. Candidates are not the driving forces in this process. To the contrary, as they have in the past, candidates compete for the support of groups that constitute parties. These “invisible primaries” determine the front-runners and shape parties' strategies.

The strength of this book is in the detail about the preconvention, invisible primary since the McGovern-Fraser reforms of the early 1970s. This extensive process of travel, endorsements, fund-raising, and signaling constitutes a period of “discussion” through which party elites sort through the candidates and often develop a consensus about the preferred standard-bearer. Addressing recent preconvention periods and nomination contests with a common set of analytical questions yields a valuable account that cannot be found in other studies.

The central analytical argument concerns the micro-level foundations of a theory of parties. Is the best foundation the electoral or career goals of office seekers or the policy goals of organized groups? Many scholars of parties, groups, and collective action will find this an unnecessary choice. In light of the problems of collective action and coordination, the choice between career-oriented politicians and policy-oriented groups seems to fade in importance. It is impossible to account for electoral parties without accounting for how politicians and groups, electoral and policy goals, interact with each other. Thus, it may be more important to specify how and why that relationship takes shape and evolves than to place politician-dominated and group-dominated activity on a scale.

For these reasons, Chapter 4 proves the most interesting. Here, the authors argue that throughout American history, presidential candidates sought to line up support among groups and local party leaders before conventions, at least to the extent that transportation and communication technology allowed. In this process, party regulars—group leaders and party elites—chose candidates whose policy stances reflected the “party” answer to the strategic challenges of keeping a party together and winning elections. In this respect, the authors insist, the McGovern-Fraser reforms did not bring a new era.

There are two key passages in the book. The first concerns the existence of the invisible primary (p. 103): “A natural question is why party members would abandon the traditional party nominating convention as the mechanism for choosing nominees and begin—probably without conscious decision—to make decisions ahead of the convention. We offer the following conjecture. Whether on the floor of the convention or in a preconvention campaign, the choice of nominee is always to some degree a bandwagon process. Candidates and a handful of key backers launch the bandwagons, but a bandwagon succeeds according to whether enough party members see it as the most promising vehicle for achieving group goals.” Apparently, party outcomes are the joint product of candidate campaigns and the interests of convention goers. Just why parties are groups, rather than groups and candidates, is not clear.

The second passage (p. 105; emphasis added): “We have noticed only one structural difference before and after the McGovern-Fraser reforms. In the prereform period, the leading candidate went from the invisible primary to the national party convention, where his supporters voted him the nomination. In the postreform period, the leading candidate goes from the invisible primary to the state-level primaries and caucuses, where his supporters help him win these public contests.” We are asked to conclude that the difference between directly voting for a candidate at a convention and helping the candidate win votes in primaries is not significant. This will be controversial.

Finally, the authors say that the strongest evidence for their argument is the observation that the longer a party is out of the White House, the more likely it is to nominate more moderate, competitive candidates. This confirms the group theory, it is argued, because group coalitions will be more likely to favor electability over their own policy goals the longer they are out of power. As a theoretical matter, it is not clear that nongroup partisans (unorganized primary electorates, ambitious politicians) would make some other calculation. At a minimum, we need an alternative hypothesis for career-oriented politicians and one is not offered at this critical juncture. As an empirical matter, for the period since World War II, the correlation between candidate extremism and years out of power is strong for Republicans (about −0.80), who have the outliers of Barry Goldwater on one end and Dwight Eisenhower on the other, but is quite modest for Democrats (about −0.50).

Two omissions deserve mention. First, Frank Sorauf, who is not cited or discussed in this book, taught us that American political parties cannot be viewed as structured, bounded, formal organizations. Rather, American party organizations have legally defined national, state, and local components, separate caucuses in legislative bodies at various levels, and supporters in the mass public with a variety of formal relationships to party organizations. Simple micro-level assumptions about the key players and their motivations are unlikely to capture the essential complexity of the ways that party officials, organized groups, politicians, and partisans interact and produce candidacies, campaigns, and outcomes. Second, David Truman, particularly in his treatise The Governmental Process (1951), made group creation and political involvement a process central to American politics and closely connected to developments in the party system. Truman, too, is ignored.

We are challenged by Cohen and his colleagues to account for both continuities and changes in American national parties and presidential nomination politics. Refocusing attention on the parties' coalitions and the preconvention process, which exhibit important continuities, is important. The theoretical value of defining parties as group coalitions is yet to be demonstrated.