At some point, instructors of introductory American politics courses must have a potentially uncomfortable discussion with their students about the paradox of voting. The instructor will likely say something like “I’m not saying that you shouldn’t vote; I’m simply saying the typical rational models cannot explain why people do.” Some cursory discussion of the “D” term will follow this with the instructor concluding, “That stands for civic duty which is sort of a vague concept. Anyway—for the test you need to know that educated people vote more.”

In her book, Meredith Rolfe attempts to solve the paradox of voting through a rational explanation of voting without vague notions of duty or unrealistic assumptions. In doing so, she challenges conventional wisdom and produces a theory that should influence theories of turnout from this point forward. The key for Rolfe’s model is to move beyond the individual. The decision to vote is not made in isolation, nor is it simply the result of an interaction between an individual and a campaign. Rather, the author views voting through the lens of a social dilemma and explains that voting is an act of cooperation. She then uses the insights from the literature on social dilemmas to argue that an individual’s decision to participate in politics through voting is a conditional decision based on the decisions of others.

Why then do people turn out to vote, according to Rolfe? Some people (likely between 10% and 20%) are unconditional cooperators. These individuals will always vote, in the same way that there are always a few people who will unconditionally contribute to a public good. For most people, however, turnout is explained by their direct or indirect connections to these unconditional cooperators. She goes further to argue that in low-salience elections, direct or indirect connections to candidates helps explain turnout.

Rolfe devotes the bulk of the text to specifying a formal model of turnout that builds off of previous models of conditional cooperation. Her main contribution is the thoroughness with which she models a social theory of voter turnout. As part of her effort to build her model, she validates her assumptions and specifies the parameters of the model using previous research, largely in behavioral economics, and some textual analysis and survey data on the “social meaning of voting.” The chapter on the social meaning of voting is perhaps necessary for advancing the larger argument, but it is the weakest in the book. The claims made in that chapter are fairly uncontroversial, and the empirical analysis is less rigorous than the book’s later empirical chapters.

The incorporation of insights from behavioral economics, however, allows Rolfe to build a more complex and predictive model of turnout than previous rational models. She uses the results of economic games as the basis for her arguments concerning conditional cooperation. This requires a conceptual leap. She asserts that turning out to vote is a social dilemma similar to contributing to a public good.

In one sense, the connection is obvious. In a standard public goods game, the socially optimal outcome is for everyone to cooperate for the public good, but individuals are better-off if they free-ride on the contributions of others. In a similar way, democracy does not work without voting, but there is no clear, direct, individual benefit to turning out. In another sense, the connection between turnout and contributing to a public good is much more tenuous. When people turn out to vote, they may all agree that they are participating in an important communal act of democratic politics, but there is disagreement about the preferred outcome.

This is potentially an important oversight. If a person is aware that other people are recycling, that person should be more likely to recycle him- or herself. If a Democrat is aware that his or her associates are turning out to vote, then that should motivate that person to vote as well. But what if those associates are turning out to vote and they are voting Republican? Then our hypothetical Democrat knows that he or she should vote, but is receiving a conflicting signal about how to vote. Many studies of contextual effects suggest that the social environment would depress turnout in this instance. Rolfe’s theory does not address this matter in any real way.

How problematic is this and the other simplifications of the model? The model is certainly incomplete, and future work should attempt to embellish it. The model is so complex, however, that simplifications are needed somewhere. This is a book for social science in general and political science more specifically, and Rolfe has chosen to emphasize the social over the political. I believe this decision is warranted. Empirical studies have shown that many voters experience some disagreement, but few individuals are in social networks dominated by members of another party. Most models and survey research on voter turnout implicitly consider voters as individuals responding to campaign stimuli, registration costs, and so on in isolation. This assumption is much more unrealistic and problematic.

The book contains two empirical chapters testing the model—one on turnout in high-salience elections and the other on turnout in low-salience elections. In these chapters, Rolfe makes the largest empirical claim of the book: The relationship between education and turnout seen in previous studies is the result of omitted variable bias. Using individual and precinct-level data, she shows that education serves as a proxy for one’s social connections. Educated individuals are more directly and indirectly connected to people who participate, and they are also better connected to candidates directly. Education, therefore, does not lower the costs of participation, as is often argued. Rather, the educated live in a social world where turnout spreads much as a musical or clothing fad would spread.

The evidence in this case is carefully considered, and the results are in line with the predictions of the theoretic model. The only problem is that the data are limited. Rolfe uses the 1985 General Social Survey and more fine-grained measures from a series of elections in Mecklenburg County, North Carolina. Limited data is a common problem with contextual studies. Much of what we know about contextual effects comes from the efforts of Robert Huckfeldt, John Sprague, and colleagues (e.g., Citizens, Politics, and Social Communication, 1995) whose work focused first solely on South Bend and then later on St. Louis and Indianapolis. The evidence presented in Voter Turnout is compelling and speaks to the need for surveys to include better contextual measures to further test all of the implications of this and other contextual models.

In the end, the importance of Voter Turnout really is that it forces researchers to be better than the typical undergraduate. All undergraduates need to know for the test is that educated people vote more frequently. Scholars and policymakers need to know why. In a recent article, Barry Burden et al. (“Election Laws, Mobilization, and Turnout: The Unanticipated Consequences of Election Reform,” American Journal of Political Science 58 [January 2014]: 95–109) note that early voting can actually decrease turnout despite supposedly lowering the costs of voting. Rolfe predicts this very result on page 3 of her book, saying that these types of reforms will backfire because they do not consider the social aspects of turnout. Too many social scientists have ignored “social” factors, which has led to bad scholarship and bad policy.