Campaign Dynamics: The Race for Governor

Chapter 4. Data and Methods

CHAPTER 4 Data and Methods In this chapter, I describe the data and methods used to evaluate predictions about both candidates and voters generated by the theory presented in chapters 2 and 3. The analysis that follows consists of two case studies along with data spanning more than one hundred gubernatorial elections held from 1982 to 1994. I have argued that as a general strategy , candidates try to induce heresthetic change in the electorate by shifting the focus of the campaign to dif ferent issues , thus altering the relative salience of various divisions among voters.As a result, a campaign is a struggle to define the nature of the election: a battle be tween the candidates to raise and/or lower the relative salience of particular issues or cleavages for voters.Thus, the interaction between the candidates themselves and with the voters that takes place during the campaign influences th behavior of both the candidates and voters. If the theory guiding this book is correct, we should observe candidates linking the themes they stress with tar get populations as the candidates develop their campaign strategies. Candidates and their staf fs should discuss the campaign in terms of setting the agenda or defining the choice for voters. Given th voters’ limitations outlined in chapter 2, candidates should focus on a limited number of issues and generally take clear stands on those issues. We should also see campaign strategies evolving as candidates learn more about their opponents and the electorate. Furthermore, if the theory presented is correct, the salience of various determinants of voting behavior should respond to the themes candidates stress. The particular characteristics that respond to the themes stressed by candidates depend on the choice of theme. For example, if candidates in one race stress the abortion issue, voter attitudes on abortion should be stronger predictors of voting behavior in that election relative to an election in which the candidates did not stress abortion. The complexity of the campaign process requires a research design that incorporates data on individual and aggregate-level factors that influence votin behavior as well as data on the themes candidates stress during elections. The 40 research design should also examine the dynamic process of campaigns. Most important, the design must isolate the interaction between the campaign themes adopted by candidates and the salience of various divisions among voters. Case Studies Does the characterization of the campaign process presented in chapters 2 and 3 fit real-world examples of statewide campaigns? Do candidates and voters re spond in the dynamic way predicted? I seek to answer these questions f rst through a detailed examination of two gubernatorial campaigns, the 1993 races in Virginia and New Jersey. Both races received some national attention, were highly competitive, and witnessed dramatic shifts in public opinion regarding candidate support over the course of the campaign. While no two elections could be considered representative of all gubernatorial elections, these two contests dramatically illustrate the dynamics of gubernatorial campaigns and the voters’ responsiveness to them. Before I discuss the data gathering process for these two cases, I will say a bit about a statistical model of campaigns implied by the theory developed in chapters 2 and 3. One can model the dynamic process of a campaign as a system of simultaneous equations that captures the interaction between the strategies of the two candidates and the salience of various cleavages among voters. The media plays the key role of linking these elements together by providing everyone involved with information about each other. The process is dynamic because the previous behavior of candidates and voters influences the current behavior of bot candidates and voters. For those who are interested, a more detailed specifica tion of such a model is presented in appendix A. Before any campaign begins, we know that certain individual-level voter characteristics, election-level factors, and national-level factors characterize the current political environment. From this initial position, candidates develop their campaign strategies. For example, the partisan makeup of a state and the role of partisanship in previous elections serves as an initial value for the po tential salience of such a cleavage. Candidate strategy is assumed to influenc the salience of such factors but not their existence. This preexisting political context can be treated as exogenous to the dynamic process of the campaign. The political context provides the bounds for the campaign. It presents specific opportunities and obstacles, and strategic can didates choose how to respond to them.The choices...