Models of voting behavior go a long way toward predicting the choices individuals make in elections. And, for practical purposes, prediction is critical, as it gives us important insight into the potential outcomes that might ensue, given certain conditions. However, research on information processing is increasingly focusing our attention more explicitly on understanding the process by which voters make decisions rather than focusing exclusively on the decisions themselves. This is exactly what Richard Lau and David Redlawsk do in How Voters Decide. This book departs from previous information processing research because not only do the authors propose a comprehensive process-oriented model of voter decision making, but they also test the various steps in the process using data gathered in an explicitly dynamic format.
In this process-oriented approach, voters gather and process campaign information in order to evaluate candidates and make electoral decisions. Evaluation, voting, and the quality of the vote decision are perceived to be a function of various elements of information processing, memory for that information, and the nature of the decision task.
The first step in the model proposes that various factors—demographic characteristics, political sophistication, and campaign features—influence the perceived nature of the decision task that individuals face in an election. In turn, the nature of the decision task influences the processing of information (depth of information search, the comparability of search across candidates, and the systematic nature of information search). Different combinations of these information elements produce what the authors term decision strategies that conform to four relatively common models of the vote decision in the political science literature: standard rational choice (Model 1), early socialization or cognitive consistency (Model 2), fast and frugal (Model 3), and bounded rationality or intuitive decision making (Model 4).
The extent to which information is processed or the decision strategy used will affect the quantity and accuracy of memory for that information. Information processing, memory, and the nature of the decision task in turn influence candidate evaluation, vote choice, and decision quality. The authors outline a series of testable implications based on this theoretical model.
To test these implications, a dynamic information environment is necessary. One of the book's important contributions is the adaptation of a standard process tracing methodology—the static information board—to the ever-changing information environment of political campaigns. In a series of experiments, the authors use their dynamic information board to create mock presidential election campaigns and trace the information that subjects choose to access about fictitious candidates and the electoral decisions of these subjects in both primary and general elections.
Rather than recite the entire array of findings from these tests, I would like to highlight a few findings suggesting that certain notions prevalent in the literature deserve further contemplation. In particular, when the authors consider the effect of individual characteristics on information processing, memory, and the quality of the vote, age is the only variable negatively related to each of these, whereas political sophistication is the only variable positively related to each of these factors. Consistently, researchers have found that both age and sophistication are positively related to greater voter turnout, yet the findings in this book suggest that the two groups of voters might be similar in their levels of turnout but not in the extent to which the vote choices they make are correct. More fully exploring such individual-level differences would certainly be a fruitful avenue for future research.
Further, this book finds that memory plays an important role in candidate evaluation, the vote choice, and the quality of the decision. While Lau and Redlawsk conclude that these findings can coexist with previous experimental research attributing a negligible role to memory (given that the latter is concerned with candidate evaluation in a “nonchoice” situation), at the very least these findings should give us pause in constructing models of both candidate evaluation and vote choice that attribute no or a limited role to memory. Additionally, these findings raise an interesting theoretical question: Why might memory affect evaluation in an experiment with an electoral context, but not in an experiment without such a choice?
The authors also examine a question raised by recent research on heuristics: Can the American public, found wanting in terms of political knowledge, make decisions as if they were fully informed about politics? By focusing on the ways that individuals can make decisions, Lau and Redlawsk argue that we can determine whether individuals vote correctly or make electoral choices that they would make had they been fully informed. Using two different measures of correct voting, the authors find that correct voting varies substantially with the number of candidates in a race—70% with only two candidates and 31% with four candidates. This finding is interesting in light of many of the criticisms of a two-party system: providing less choice. These findings suggest that one might need to trade off choice for quality.
Perhaps even more significant are the findings that voters using a rational decision strategy recall less information and make worse decisions than voters using any of the other three strategies. In fact, the rational strategy performs worse than chance under the more difficult electoral conditions (four candidates) and no better than chance under simpler conditions (two candidates). Why does the rational strategy perform so poorly? The authors argue that in a presidential election, voters are overwhelmed by information and incapable of processing the information according to a rational model in a way that would yield a correct choice. Could it be that the rational model is simply too difficult for most voters to use in such a context? In part, the authors find this to be true as Model 1 sophisticates vote correctly more often than the baseline, whereas Model 1 novices do substantially worse. Such findings run counter to previous research suggesting that a rational strategy should be useful in exactly those situations that the authors find it to be least useful—complex decision-making environments. An important question remains though: Can rational strategies prove more useful to voters in less information-rich elections like House and Senate races with and without incumbents?
In sum, How Voters Decide makes three critical contributions: proposing a process-oriented framework, testing this framework using a dynamic information environment, and outlining a variety of findings that raise critical questions for future research.
