An important and ambitious empirical study of policymaking in the (K–12) school districts of America, this book presents statistical support for noteworthy and sometimes surprising conclusions about the functioning of democracy at this level. The study is important because it raises and answers theoretically valuable questions about democracy in the organization and functions of the public schools. It is ambitious because the data and analytical requirements upon which these questions and answers depend are substantial.

In an empirical contribution to democratic theory, the authors take an essentially outcomes view. They see democracy as “policy responsiveness,” which, in their case, means the correspondence between public opinion on school spending levels and the actual level of spending in the district. Among the most interesting conclusions of the book are the roles of the different institutional structures across American school districts and their differential impact on policy responsiveness.

American school districts use a variety of institutional forms for making policy, including decisions on taxing and spending. They vary from direct democracy, such as New England town meetings and voter initiatives for tax and spending decisions; to representative democracy through elected school board decisions on these issues, in some jurisdictions combined with citizen referenda, in others, not; to school boards not directly elected but appointed by higher authorities, such as city, county, or state officers.

Most observers would view this array as one of a declining level of democratic forms, with town meetings as the most democratic and appointed boards as the least. Consequently, one would expect policy responsiveness to decline across the array. Yet the findings are the reverse. Policy responsiveness is greatest when decision makers are appointed and is lowest in the New England town meetings. The authors' explanation is, in effect, that indirect democracy works: Public officials enact general citizen preferences more closely than do low-turnout town meetings and elections.

Similarly, the function of what the authors call “substantive representation” is consistent with this explanation. They use that term to mean, in effect, policy responsiveness to African Americans. It differs from “descriptive representation,” which means whether the school board membership approximates the African American proportion in the district. In the South, the move toward elections and away from at-large elections increased descriptive representation, but the fundamental finding is that appointed school boards were the most representative of, and had the greatest level of, policy responsiveness to African Americans.

The findings about institutions and policy responsiveness have implications for public choice theory. Scholars from that perspective tend to hold that representative structures, such as school boards, are one or more steps distant from the preferences of the “median voter” and that some form of direct democracy would come closer. In policy terms, these scholars usually hold that this means that voter-isolated school boards and appointed school superintendents (and other bureaucrats) spend more on the schools than the average voter prefers. Yet as the authors point out, there has been heretofore no empirical support for that conclusion, and the authors' conclusion holds otherwise.

Other interesting findings include the impact of teachers' unions and the aged. The former, like school bureaucrats, have been accused of pushing school spending beyond citizen preferences, and the latter, the reverse.

The authors show that teachers' unions that lobby for increased school spending are effective at state and local levels, depending upon their strength and memberships at those levels. Nevertheless, they conclude that their effectiveness is just as often consistent with public opinion as not. Often the unions and the citizenry are united in advocating increased school spending.

The “Gray Peril” hypothesis is that older citizens in the community will try to hold down school spending as they no longer have a direct interest in children's education. A competing hypothesis is that they do have an interest in maintaining property values, which depend in part on good neighborhood schools. The authors say that the aged have differential effects on school spending. Those who are long-term residents support spending on the schools as they identify with the community, whereas newcomers of the same age group tend to depress school spending. Pity the Florida children.

In both of these findings and others, the authors make use of interaction effects to estimate the differential impact of groups on policy. This analytical approach is refreshing in a book of this nature; it does not limit its analysis to additive effects.

The analytical ambitions of the study include the need for public opinion measures from the 14 thousand school districts in America. No one can afford such a study, and the authors rely on a technique initiated in political science by the Warren Miller and Donald Stokes 1963 study of constituency opinion and congressional voting. They used a national sample broken down to congressional districts to measure opinion. A small industry grew up around this method, continuing with the present book. As a matter of full disclosure, I (with Donald J. McCrone) participated in a small way in an early (1966) and much-criticized part of this industry. (Indeed, a subsidiary industry of such criticism shortly developed.)

An issue that has not been well aired in these critiques is whether these opinion measures come from representative samples. In the present book, the authors add cases by using the opinion question on respondent preference for higher, lower, or the same level of spending on the schools that employs the General Social Survey (GSS) data from the 1985 to 2002 surveys. Aggregating, they achieve 26,000 respondents, enough to satisfy sampling error requirements. Yet sampling error estimates assume random sampling. We assume that the GSS meets this requirement for the nation as a whole. Because of the stratifying and clustering factors in a national sample, the conclusion that a sample—no matter how large—is representative of a component unit smaller than the nation, such as a state, or a congressional or school, is problematic.

The authors call this a part of their “small policy inference” approach. The approach uses school districts, ranked by opinion regressed on measures of demographic and geographic characteristics. The results of this regression form the independent variable, opinion, which is the basis of another regression in which expenditures are the dependent variable. The discussion of this analysis would be improved by sensitivity to the econometric and political science literature on the “error in variables” problem.

Michael Berkman and Eric Plutzer have provided us with a careful analysis of important issues that inform democratic theory, empirical analysis, public policy, and education in the United States. It needs to be on reading lists in all of those fields in political science and the other social sciences.