In this book, William Claggett and Byron Shafer advance two broad claims about postwar politics in the United States: first, that from the end of World War II to the present, public opinion on vital matters of policy has been well structured; and second, that from Harry Truman to Barack Obama, voters have acted on their policy preferences when choosing a president.

The authors launch their enterprise by asserting that postwar politics in the United States has taken place primarily within four policy domains: social welfare, foreign affairs, race, and culture. This is a reasonable claim and a defensible point of departure (though referring to it as a theoretical achievement, as Claggett and Shafer do, seems strange). Four chapters follow, each devoted to one of the policy domains.

These chapters all open with a page or two of historical background—the emergence of the United States from the Great Depression and World War II, in the case of the chapter on social welfare, for example, or the rising of the Civil Rights movement in the case of the chapter on race—and then get down to business. First, every policy question potentially pertinent to a domain appearing in any of the studies carried out by the National Election Study (NES) from 1948 to 2004 is identified. Then, each question is carefully reviewed. Many are set aside, either because upon close inspection they seem peripheral to the domain or because they appear to belong to more than a single domain. For example, the long-running NES question on whether or not the federal government should make every possible effort to improve the standing of black Americans is discarded because the question is about social welfare and it is about race. Also abandoned are questions on trade policy, controlling urban unrest, federal spending on AIDS research, and protecting the interests of working women. Removing such policies from analysis may help to ensure the purity of each policy domain, but in the process, a fair amount of what is interesting in American politics—politics that takes place within the intersection of established domains—disappears from sight. That said, the work here is careful and thorough, and will be of value to others intent on exploring the issue preferences of the American public.

Policy questions that survive this screening are then factor-analyzed (when items are sufficient in number to support such a procedure), one election study at a time, first to see whether opinions within a domain are correlated, and then, using confirmatory methods, to see whether opinions within a domain are correlated in predicted ways. Claggett and Shafer generally find that public opinion within a policy domain consists of distinct but correlated clusters. Within the domain of social welfare, for example, Americans distinguish among education, child care, social insurance, unemployment benefits, and income-based transfer programs. These represent related but distinct aspects of the American welfare state.

The domain of culture—the collection of issues having to do with how the American people should live their lives—makes this point more emphatically. According to Claggett and Shafer, the cultural domain of policy encompasses the changing place of women in American society, matters of crime and punishment, issues arising from immigration and national integration, gay rights, abortion, religion and public life, and more. Do these various and sundry issues really belong together? It is hard to say for sure, because the full set of measures necessary to provide a definitive answer never appears on a single study. But when items appear on an NES study in numbers sufficient to allow correlational and confirmatory factor analysis, results suggest (to me at least) that Americans do not think about these issues in the same way—that women's proper place and immigration and gay rights and so on constitute largely separate issue clusters.

As for consistency across the four domains, Claggett and Shafer find virtually none. Views on national security have little to do with opinions on social welfare, which in turn have little to do with positions taken on abortion or gay rights, and so on. The big picture here, then, is modest and for the most part intelligible consistency within domains—more in some domains (social welfare), less in others (culture)—and no consistency across domains. To students of public opinion, this is an utterly familiar pattern. For 50 years now, we have known that most citizens come to politics without a global point of view, and that public opinion on policy consists of a proliferation of relatively narrow and largely unconnected concerns.

The authors interpret their results differently, as of course they are free to do. They offer their results as a radical challenge to what they take to be the conventional wisdom and as a sharp rebuke to those analysts of public opinion they see as bent on undermining the foundations of democratic politics. Such analysts are unnamed and never cited, but their “prejudices” are conveniently listed on page 275. The list begins with “The public generally lacks policy preferences” and concludes with “So, in the end, American mass politics is remarkably free of programmatic substance.” In between come a set of sweeping propositions to which no analyst I know professionally would put their name.

The second part of the book takes up the practical effect of opinion on policy, as revealed in presidential voting. For this purpose, Claggett and Shafer want to claim that they have created adequate and comparable measures of public opinion on each of the four grand questions that have been at the core of American politics in the postwar period. This would allow them to compare the role of opinion in presidential elections over time and across policy domains.

As a practical matter, complexity within policy domains makes trouble for this aspiration. If opinion within each domain were reducible to a single dimension, Claggett and Shafer could proceed in a relatively straightforward fashion to their vote analysis. But opinion is not reducible to a single dimension. Moreover, the policy questions within the four domains vary from one NES to the next. From the beginning of the NES series to the end, they vary dramatically. Acknowledging these results, and facing this predicament, the authors argue that, nevertheless, comparable measures can be cobbled together, and so proceed to estimate the impact of policy views on presidential voting—what they call the “pure politics of policy choice.”

Presidential elections are taken up from 1948 to 2004. Each election is considered separately. In each case, vote choice is predicted by views on policy—and by views on policy alone. Thus in operational terms, the pure politics of policy choice appears to mean that elements of vote choice other than policy—partisanship, economic conditions, war and peace, domestic tranquility, especially attractive or especially repellant candidates—are simply set aside.

The authors also estimate their model within partisan groups, which seems a clear improvement in specification. They find that the effects of policy opinion on the vote are much more pronounced among Independents than among Democrats or Republicans. Of the four policy domains, opinion on social welfare policy is generally the most important. Moreover, while the effect of the social welfare domain is quite stable across elections, the impact due to opinion in other policy domains is episodic: National security shows up in the Reagan elections; race peaks in 1964 at the epicenter of the country's twentieth-century racial emergency; and the cultural domain emerges as a significant factor only in recent contests. As such, these findings are broadly consistent with the well-established proposition that the magnitude of issue voting depends upon the extent to which rival candidates offer voters a clear choice.

Taken all around, Claggett and Shafer present an abundance of fascinating results. As always, readers get to decide what the results mean—in this case, whether the findings presented in The American Public Mind upend conventional wisdom or largely confirm it.