Information and Elections

9. Campaigns and Uncertainty

CHAPTER 9 Campaigns and Uncertainty Political campaigns are designedly made into emotional orgies which endeavor to distract attention from the real issues involved. -James Harvey Robinson, The Human Comedy, 1937 The politician is. . . trained in the art of inexactitude. His words tend to be blunt or rounded, because ifthey have a cutting edge they may later return to wound him. Edward R. Murrow, London, October 19, 1959 The focus of analysis so far in this book has been on the individual voter. Examining microlevel decision making has yielded a number of insights into how information influences voter choice. I have presented a considerable amount of evidence which documents the variation across the voting public in their uncertainty about the policy positions of presidential candidates, and how differences in relative uncertainty affect voter preferences and the criteria upon which they base their choices. However, as noted in chapter 3, perceptual uncertainty is rooted in both the voters' own disincentives to gather and process costly information and in the candidates' incentives to disseminate ambiguous information about their policy positions. I concentrated in the previous chapters on variables related to voter information costs. Here, in the last two chapters of the book, I turn my attention to uncertainty which is induced in voter perceptions by the campaign (which I termed exogenous uncertainty in chapter 3). Analysis of candidate-induced uncertainty, then, implies a macrolevel focus on the campaigns conducted by the candidates. This is a difficult task, since the period of analysis covered by this study only contains four elections and nine different campaigns.1 Methodologically, there are many campaign-level I. In this chapter I do not consider the 1992 election. The operationaiization of the uncer157 158 Information and Elections variables which could account for variations in candidate-induced uncertainty, and modeling the independent effects of so many variables is not possible. Although the discussion in this chapter is more qualitative than quantitative, the evidence is compelling. This chapter begins by recounting briefly the main findings thus far in the study of voter uncertainty. Then, I present data on the aggregate distribution of voter uncertainty across the four elections. The chapter turns to a discussion of the patterns apparent in these aggregate distributions, and I first present evidence of some of the campaign-specific factors that do not account for the apparent patterns. I then argue that the distributions of voter uncertainty across the four elections are actually part of a general process of learning about political candidates. Explaining Aggregate Uncertainty In chapters 7 and 8, I presented tests of three hypotheses. Two of the hypotheses received strong and consistent empirical support: that the greater the voter's uncertainty about a presidential candidate, the lower their evaluations of that candidate, and the greater this policy uncertainty, the lesser the voter's ability to employ their issue preferences in their decision making. The third hypothesis, which asserted that voters uncertain of the policy positions of the presidential candidates would be more likely to utilize nonpolicy (partisan) information in their decisions, received only mixed support. However, there was considerable variation across the four elections in the apparent magnitude of these effects of policy uncertainty on voter choice. Here, I recount this variation and argue that the patterns shown in these previous results are related to the aggregate levels of uncertainty in these campaigns . First, regarding the direct effects of voter uncertainty on candidate preference, I demonstrated that these effects were the greatest in the 1976 and 1988 elections, followed by the 1980 election. The 1984 election showed the least influence of voter uncertainty on candidate preferences. Next, I presented evidence about the extent of issue voting in each of these elections. The effect of policy issues was greatest in the 1980 and 1988 elections, closely followed by the 1976 race. Policy issues were shown to have had the least effect in the 1984 election. Thirdly, regarding the effects of uncertainty on issue voting, the interactive models of chapter 7 demonstrated the largest effects of uncertainty in the 1976 and 1980 elections. The 1988 election followed, while in the 1984 election uncertainty had the least interactive effect. Also, the interactive models protainty variable is different enough for the 1992 election (since it is based only on ideology) to make it difficult to compare directly to the other elections. Campaigns and Uncertainty 159 vided support for the hypothesis regarding uncertainty and the use of...