In lieu of an abstract, here is a brief excerpt of the content:

CHAPTER 9 Voter Response to Gubernatorial Campaigns, 1982–1992 Chapters 7 and 8 demonstrated that voters in gubernatorial elections responded in 1990 to appeals made on the abortion issue and in 1994 to appeals based on evaluations of President Clinton. These two chapters provided support for the theory of electoral politics outlined in chapters 2 and 3 by showing that the salience of various demographic cleavages, as well as the salience of opinions on abortion and presidential evaluations, responded to what candidates said during their campaigns.This final empirical chapter expands the analysis to in corporate more issues, more cleavages, and more gubernatorial campaigns. In this chapter, I explore how the salience to voters of gender , race, religion , income, and partisanship responded to campaign themes in seventy-one gubernatorial elections held between 1982 and 1992. 1 This is not a random sample of elections. Lar ge states and competitive races tend to be somewhat overrepresented. Still, races from thirty-seven states from every region in the country including usable responses from more than ninety-three thousand voters are included in this chapter . To my knowledge, this chapter analyzes the largest single pool of voters in gubernatorial elections ever assembled. The theory described in chapters 2 and 3 predicts that candidates strive to induce heresthetic change as a general campaign strategy. This chapter tests that prediction under more general circumstances. To begin, table 9.1 presents a list of the seventy-one elections included in this analysis along with the main issues stressed by the two main candidates. I coded up to four issue-based themes for each candidate, although as table 9.1 makes clear, most of the campaigns covered in this analysis focused on fewer than four main issues. The process of coding the content of campaign themes was described in detail in chapter 4. The theory of electoral politics guiding this book assumes that the salience of individual-level characteristics to voters varies across elections. Of course, the key assumption is that this variation results in part from the themes candi148 dates stress during their campaigns. However , before estimating a model designed to explain this variation in how voters behave across elections, I firs demonstrate the existence of that variation. For each election, I estimate a logit model predicting the probability of voting for the Democratic candidate for governor. Table 9.2 presents the logit coefficients operating on the independent variables gende, race, income, party identification, and Catholicism for each election. Whether a voter is female, African-American, Catholic, a Democrat, or a Republican is measured using dummy variables for each, coded a 1 if the voter has that characteristic and a 0 otherwise. Income refers to the total income of the respondent’s family and is coded on a 1 to 5 scale, with higher values indicating higher incomes. The dependent variable for each model is whether the respondent reported voting for the Democratic candidate for governor. The dichotomous nature of the dependent variables suggests using logit to estimate the models.2 The findings presented in table 9.2 demonstrate that the salience of thes factors varied substantially across these elections. Both partisanship variables achieved statistical significance in all but one case, and all are in the expecte direction. Still, the magnitude of the logit coef ficients operating on the tw party-identification variables varied from election to election from those wit an absolute value of less than .7 to several with an absolute value above 2.0. The four demographic variables show equal volatility , occasionally achieving statistically significant relationships in the opposite direction fro what would normally be expected. Comparing coef ficient estimates from mod els applied to different data sets should be done with caution. However , table 9.2 does illustrate in a straightforward manner that the importance of particular individual-level characteristics in predicting voting behavior varied across these elections. Table 9.3 begins the search for the role played by campaign themes in structuring gubernatorial voting behavior by laying out the expected relationships between several campaign themes and sociodemographic characteristics. It shows that I expect the gender cleavage to respond when candidates stress environmental issues, crime, and/or abortion as part of their campaign themes. The gap between blacks and nonblacks should respond to campaign themes targeting education and/or crime issues. The salience of personal family income should respond to candidates’appeals based on income differences among voters (i.e., “I am for working people; my opponent is for the rich”). Chapter...

Share