Peter Nardulli's book blends syntheses of previous research and new empirical analysis to address the question of the capacity of American voters to hold political leaders accountable through elections—to achieve, that is, “popular efficacy.” The book begins by interpreting previous scholarly research as depicting voters as “manageable fools” because of their limited cognitive abilities for political evaluation (Chapter 2) and their habitual reliance on party identifications to dictate their vote choices (Chapter 4). Given this, why do politicians pay attention to public concerns? The original empirical analysis of the book answers this question. It shows how variations in and deviations from the normal vote are related to exogenous forces beyond the control of politicians, rather than habitual party loyalties or political strategies. The realization that their fortunes are tied to forces beyond their control, in turn, induces strategic politicians to be responsive to the public's core concerns.

The book's central empirical analyses relate local (county/city) presidential results from 1828 to 2000 to various demographic, performance, and partisan indicators—measured at the local level where available and at higher levels of aggregation where local measures are unavailable. Multielection moving averages of presidential results determine each electoral unit's normal (or party) vote—the aggregate-data operationalization of Philip Converse's well-known concept. Changes in this normal vote and deviations from it in vote results suggest the presence of exogenous (i.e., nonparty) influences. The development of this rich data archive is a valuable contribution itself, but it is how these data are employed to address important democratic-theory questions about voting behavior that constitutes the major contribution of Popular Efficacy in the Democratic Era.

The impact of a work typically depends on its new or unexpected results, and this book provides a number of them. Chapter 6 compares the “machine model” of steadfast partisan loyalties against the “pluralist model” of partisan loyalties contingent upon a party's representation of social groups. The machine model is supported only at the tail end of the nineteenth century and into the early twentieth century. Even during this period and throughout the entire 1872–2000 period analyzed, the pluralist model fits voting patterns much better. Rather than serving as a reservoir that party elites can draw upon at any time, Nardulli concludes, partisan loyalties must be earned continuously by the parties' responsiveness to social groups. Moreover, with social change, the social bases of the parties themselves change—sharply (critically) in 1932 and 1952, more slowly (secularly) at the turn of the twentieth century and in response to the Civil War. So important are these group bases of voting in the book's macro-level account that they should be receiving more attention in voting behavior specialists' micro-level studies.

Chapter 7 systematically examines electoral changes that denote various realignment-triggered electoral eras. While the analysis confirms the familiar critical elections of 1856, 1896, and 1932, it adds 1836 and 1952 as turning-point contests. Nardulli fixes 1952 as the start of the “Civil Rights era” in which white southerners began to move into the Republican Party. (Even if this migration began in 1952, though, the label ignores the class, not race, basis of the pre-1960s movement that other scholars have demonstrated.) He also finds that in the last quarter of the nineteenth century, when political parties are thought to have had their greatest hold on voters, most locales experienced substantial changes in presidential voting patterns.

Chapter 8 turns to “electoral perturbations” (i.e., temporary deviations) from the normal voting patterns and their relationship to exogenous factors (economic conditions, war and peace, crime) versus endogenous factors that parties control (home state, war hero, and incumbent candidates, and party control of government). Although incorporating such variables into the analysis often required a shift to state and national aggregates, the results are worth the price. Some findings are familiar: Economic conditions are powerful influences on electoral results, and presiding over wars erodes the electoral strength of the incumbent party. Other results are surprising: Crime rates have little effect on Democrats, while increasing crime rates hurt incumbent Republicans. The often-touted home state, war hero, and incumbent advantages turn out not to be advantages at all. Monopoly control of all branches of government over time depresses support for the majority party. That exogenous forces influence voting results more than those easily manipulated by strategic elites, Nardulli concludes, makes voters less “manageable” and politicians more attuned to popular wishes.

Chapter 9 turns to the distribution of presidential election results over time across various levels of competitiveness and fidelity to party (normal vs. deviating vs. endorsement elections). While competitiveness is very rare at the local level, it increases at each higher level of aggregation—and is especially pronounced in the states before the Civil War and nationwide through 1884, then occasionally after World War II. Surprisingly, normal elections were far more common prior to the Progressive era and have been rare ever since. In this sense, the Progressive reforms seem to have realized their goal of increasing accountability.

Addressing questions about voting behavior from the standpoint of aggregate election results is a challenging enterprise. Nardulli deserves compliments for his effort in building a valuable subnational election data archive, his caution in drawing inferences from aggregate data, his creativity in devising aggregate indicators of voting patterns and their possible determinants, his methodological sophistication (including his ingenious estimates of electoral college impacts), and the novel results his analysis has produced. This short review cannot do justice to the book's contributions.

At the same time, one experiences the uneasiness about what the indicators are actually measuring that is common in aggregate analysis. For example, a normal vote estimated solely from election results for such a visible office as the presidency is a questionable indicator of the underlying partisanship of an electorate, especially one with many independents and varying rates of turnout. Nor is it obvious that party control of the legislature is a useful surrogate for political machine strength, that nationally measured indicators can capture the effects of economic conditions adequately, or how crime rates should affect the vote prior to modern times. In my judgment, Nardulli also overstates the negativity of the survey evidence on voter capabilities, neglecting the “cognitive miser” characteristics of decision making by both masses and elites and understating the roots of party identifications in issue- and group-oriented responses to political competition. Finally, the identification of 1952 as a turning point election begs for further exploration, especially given its position as one of two foundational elections for the Michigan model.

Despite these problems, this as an important book for scholars of voting behavior, American politics, and democratic theory. One of its great strengths is how well the empirical results are brought to bear upon democratic theory questions. Some may be uncomfortable that accountability, in the end, depends upon politicians' fear of electoral risks from forces beyond their control and, especially, that these forces are based more on performance than policy. Nonetheless, by empirically demonstrating these risks, their sources, and their consequences, Popular Efficacy in the Democratic Era has added valuably to our understanding of political accountability in America.