Information and Elections

6. The Causes of Uncertainty

CHAPTER 6 The Causes of Uncertainty I know of no safe depository of the ultimate powers of the society but the people themselves; and if we think them not enlightened enough to exercise their control with a wholesome discretion, the remedy is not to take itfrom them, but to inform their discretion by education. -Thomas Jefferson, Letter to William Charles Jarvis, September 28, 1820 In this chapter, I present the first set of empirical results from the two-stage uncertainty and evaluations models-the models for voter uncertainty of the presidential candidates, 1976-92. The purpose of these models is twofold. The primary purpose of these models of voter uncertainty is substantive, since it is imperative to understand the processes at the individual level which drive this uncertainty. Basically, these models answer substantive questions about the determinants of voter uncertainty; that is, which voters are more uncertain than others? By answering this question, I can move to the next and more important question-what might be done to better inform voters? Also, these models justify the measurement approach to voter uncertainty which is the basis for the empirical work in this book. Recall that the measure for voter uncertainty is the variation in voter placement of each candidate across a number of issues, representing the difference between the voter's perception of the candidate's position and the candidate's actual position across different issues. Here, by demonstrating that this uncertainty measure is correlated with a set of independent variables across voters in theoretically expected patterns, I increase confidence that this measure taps the uncertainty voters actually possess about the policy stands of presidential candidates (Alvarez and Franklin 1994; Cook and Campbell 1979). A good deal of the theoretical discussion of individual-level determinants of uncertainty is in chapter 4. The first section of this chapter returns to that 93 94 Information and Elections discussion and presents the theoretical expectations for the individual level determinants of uncertainty. The second section gives the models and discusses the results. Appendix C contains details of the operationalizations of the independent variables in the models, and the reduced-form model results for the candidate preference instruments. Theoretical Expectations In chapter 4, I argued that three factors are related to voter uncertainty about the policy positions taken by political figures: the flow of information in the political environment, the voter's personal information costs, and the voter's attachment to the political world and their exposure to information aboutpolitics . Since the first, information flow in the presidential campaign, is a dynamic concept, it will be impossible to include in these static models of voter uncertainty , estimated with one set of survey data (typically from October) from each election year. The influence of the campaign's information dynamics will be discussed in great detail in subsequent chapters. However, the latter two factors can be operationalized from available survey data in each of these presidential elections. I discuss these two factors in turn. As both those working in the rational choice and political cognition fields have discussed in the literature, the voter bears considerable costs in gathering , analyzing, and evaluating information. This leads me to expect that those voters to whom the costs of obtaining, processing, and recalling information about presidential candidates are the lowest should be systematically more certain about the policy positions of the presidential candidates. Correspondingly, those who have greater information costs should be more uncertain of the policy positions of the presidential candidates. A number of variables can be operationalized to account for differential information costs across voters. The first measure of information costs is the voter's education, since the better educated the voter, the better their ability to utilize information about the candidate's policy positions. The second variable is the voter's store of objective political information, which is a direct representation of how much a voter knows about political affairs (Zaller 1989). Thus the more politically informed and the better educated the voter, the less uncertain I expect him or her to be of candidate policy positions. Two other variables are included in these models to account for additional social and demographic factors that might lead some voters to be more uncertain about candidate policy positions than others. These are the race and gender of the voter. Given the impediments that have often been placed in the paths of both minorities and women in American society and political life, it is likely that these two social groups might be...