Television and Political Advertising

Volume I: Psychological Processes

Frank Biocca

• Publisher: Routledge
• Year: 2013
• eBook ISBN: 9781135437299
• Topic: Languages & Linguistics
• Subtopic: Communication Studies
• Index: Languages & Linguistics

This volume represents one of the first major scholarly effort to unravel the psychological and symbolic processing of political advertising. Utilizing survey, experimental, qualitative, and semiotic methodologies to study this phenomenon, the contributors to Television and Political Advertising trace how political ads help to interpret the psychological reality of the presidential campaign in the minds of millions of voters. A product of the National Political Advertising Research Project, this interdisciplinary effort is valuable to researchers in advertising, communication, and consumer psychology since it helps define future work on the relationship between television, politics, and the mind of the voter.

This volume, Television and Political Advertising: Psychological Processes, is the first of two, and covers such topics as Models and Theories for Viewing Political Television; Psychological Processing of Issues, Images, and Form; Differential Processing of Positive and Negative Advertising; and The Psychological Contexts of Processing.

I

Viewing Political Television: Models and Theories

1

In Search of the Model Model: Political Science Versus Political Advertising Perspectives on Voter Decision Making

John Boiney

David L. Paletz

Duke University

Voting is at the heart of democracy. Naturally and inevitably, various political thinkers have long labored to identify and explain the crucial determinants of Americans’ voting decisions. From the Columbia school in 1940, down to the present, social scientists have sought not only to unravel but also to model the processes by which voters make their choices.

Certain political activists have been struggling with the same task even longer, attempting to influence election outcomes by using the media to communicate with the voters. Known nowadays as campaign consultants or perhaps more colloquially as “media gurus,” they too possess, at least implicitly, models of the voters’ decision-making processes.

Surprisingly, the models of these two groups have never been compared, nor their differences and similarities mapped and explained. It is to such a comparative task that this chapter is devoted.

To facilitate comparison of the models, we need to establish common terms. There are five elements found in most political science models that are of continuing theoretical importance. These are partisan identification, candidate issue positions, candidate image, voter group membership, and retrospective voting. These have also been explored, although more indirectly and sporadically, in the political advertising literature. We deploy them to organize analysis throughout the chapter and directly to compare the political science to the political advertising models.

The chapter is divided into three sections. First, we review the major political science models, drawing conclusions about the current state of the literature regarding each of the five factors. Second, we survey the political advertising literature that, although extensive, does not attempt specifically to outline a model of the process by which voters decide for whom to vote. Nonetheless, some of the work is related closely enough to our task that a review is appropriate and conclusions possible.

Finally, having revealed a number of gaps in the political advertising literature, we attempt to fill some of them in with our own original research. We present and discuss the findings of our exploratory study coding the content of 196 political advertisements in terms of the five elements. In effect, we outline the current model of voter choice held implicitly by political consultants as represented in the ads they devise.

The chapter culminates with a discussion of how this consultants’ model compares to that of political science. A modicum of modesty tempers our conclusions, but does not deter us from suggesting future lines of research.

Political Science

There are four variables common to most models of voter choice. First is some form of partisanship, usually understood as partisan identification or party ID. The term originated with The American Voter (Campbell, Converse, Miller, & Stokes, 1960) and was defined as a psychological attachment to one of the two major parties. Since then, the understanding of party ID has changed considerably, with concomitant change in the variable’s centrality to voter decision making.

The second major factor is voter assessment of candidate—or party—issue positions. The common assumption regarding what we simply call issues is that voters care about what candidates or parties have done and propose to do about particular policy issues, certain economic and social problems, subjects of concern, and so forth.

A third variable we dub candidate image or simply images. This stands for the attitudes and feelings voters have toward the individuals running in a given race. Measurement of this element has varied widely over time, ranging from the open-ended likes/dislikes questions of the Michigan team to Rahn, Aldrich, Borgida, and Sullivan (1990) effort to identify three particular factors as comprising overall candidate image.

Fourth is what we call groups. Most prominently associated with the Columbia studies of the 1940s, this element refers to the tendency for some individuals to define themselves as members of a group—such as racial, religious, or economic—and to be influenced in their voting by that membership.

In addition to these four variables is the idea of retrospective voting. This refers to the argument that voters judge the incumbent alone (Key, 1966), or incumbent and challenger (Downs, 1957; Fiorina, 1981) by past events and actions with which these candidates are identified or associated.

The form and importance of the four variables and retrospective voting have varied considerably from the earliest voter models to the present. They can therefore be used to organize a roughly chronological review of the major models. Beginning with the Columbia school’s 1940 and 1948 studies and concluding with the most recent published effort (Marcus, 1988), we sketch the development and changing status of the five elements.

The Major Models

One of the earliest attempts to model voter behavior came out of the Columbia school. Lazarsfeld, Berelson, and Gaudet (1948) conducted a panel study of 6,000 voters in Erie County, Ohio from May to November 1940.

The analysis presented voting as very stable because voters erect a “sort of protective screen” to shield out any information that might conflict with their pre-set decision. One of the most basic and primary forms of that screening is an individual’s partisanship. A related major finding is that people vote “in groups,” meaning they tend to follow those with whom they work, play, worship, and so on. Thus, a person’s demographic characteristics provide a stable predictor of the vote. The researchers linked the two findings in an “index of partisan predisposition,” constructed from such demographic dimensions as religion, residence, and socioeconomic status (SES).

Thus, at this early stage of voting study research, partisanship was an important variable not only because of its screening function, but because the Columbia scholars characterized voting as a choice between parties, not candidates. However, group membership is ultimately the central variable: It produces partisanship.

Although Downs (1957) also constructed his model in terms of the parties, it is quite different from that of the Columbia scholars. He posited a rational voter who compares what he or she believes are two programmatically consistent parties in an effort to maximize the utility he or she would receive were each in office. The voter weighs the incumbent’s performance against what the challenger would have done, deriving an “expected party differential” (p. 40). If it is positive the individual votes for the incumbent, if negative, for the challenger.

Thus, although Downs’ model too represents a partisan choice, it says nothing about partisanship. And an individual unmoved by group ties makes that choice. He or she uses the past in rational, issue-based terms to choose for the future.

The American Voter (Campbell et al. 1960) represents the pinnacle of the partisan approach to voting. The Michigan researchers introduced the concept with which they have become identified: partisan identification. This concept is significant for three reasons. First, it is explicitly psychological, a decided departure from both the Columbia group and Downs. Second the Michigan men determined it to be a durable and stable measure. Third, they assumed party ID represented a “perfect distillation of all events in the individual’s life history that have borne upon the way in which he relates himself to a political party” (p. 34). Here, then, was a single, easily measured variable that seemed to allow scholars to tap the voter’s mind.

The Michigan scholars also initiated systematic inquiry into the influence of attitudes toward the candidates. They used questions tapping likes and dislikes of parties and candidates, and categorized responses across six dimensions of “partisan attitudes.” This approach heralded the coming of a new and significant variable—candidate evaluation. Although voter choice is still primarily a function of partisanship, it is no longer purely a choice between parties. The questions in the likes–dislikes survey place equal emphasis on attitudes toward candidates as toward the parties.

Kelley and Mirer (1978) extended this reasoning. They argued that voting is a simple act with a single rule: Voters tally their likes and dislikes for each party and candidate, then vote for the candidate with the highest positive number. If the tallies are equal, the voter follows his or her party identification. Applying this simple rule to SRC data for presidential elections from 1952 through 1968, the authors were able to account for voting and election outcomes much more successfully than any previous model.

It is important to note that for both the Michigan studies and Kelley and Mirer’s work, issues can exert an influence. Because the original SRC questions were open-ended, the final attitudes yielded by coding ranged widely in content and sophistication. Some represented simple gut reactions, whereas others spoke specifically of stances taken on particular issues. Attitudes may, as Kelley and Mirer expressed it, “implicate issues of policy” (p. 573).

In 1966, Goldberg brought together for the first time all four major variables in one schematic model of voter choice. His recursive model had a series of causal paths all leading toward the vote. At the head of each path was a variable roughly corresponding to one of the four we have adumbrated. Party ID was once more the central element standing at the crossroads for all the other variables. It exerted the most substantial direct impact on voting, as well as considerable indirect impact via issues and candidate evaluation. Goldberg also found candidate evaluations exerted a much stronger influence than issues.

Goldberg’s model was the most sophisticated yet because demographics affect partisanship, and party ID affects issue positions and candidate evaluations. Pomper (1975) extended that model by adding two crucial links. He suggested that issue positions are affected not only by party ID but by the voter’s partisan predispositions (SES). More importantly, he argued that issues affect candidate evaluation. These links have two implications: a weakening of the impact of party ID, and an increase in the impact of issues.

Using data from several presidential elections, Pomper found the direct and indirect effect of party ID on the decline, issues on the rise, and the independent influence of candidate evaluations also increasing. Because transmission of party ID seemed to be weakening, Pomper concluded that voting was not a group dynamic, not a “dependent act” (pp. 200–202).

The recursive model, because it suggests that causation runs only one way, was problematic. Political scientists asked, for example, how realistic it was to suppose that issues could affect candidate evaluations without the reverse also occurring. Jackson (1975) responded with the first nonrecursive model. He simultaneously related the endogenous variables of people’s issue preferences, their evaluations of the parties’ positions, and the strength of their party identifications to one another. Using data from the 1964 presidential election, Jackson determined that party identification is derived not so much from social characteristics as from issue positions. He thus became the first researcher to offer evidence for the suggestion that party ID can change as easily as issue positions.

Although issues took on renewed prominence in this nonrecursive model, party ID remained the central variable. Jackson’s model used four simultaneous equations, each of which was either based on or determined party identification. Party ID is an integral element of issue positions, party evaluations, and the vote. The same cannot be said for any other variable. Noteworthy, too, is the fact that candidate evaluations are nowhere in sight.

Page and Jones (1984) followed Jackson with their own nonrecursive model. Three variables—current party attachment, comparative policy distances, and comparative candidate evaluations—are endogenous. Among the exogenous variables are demographic characteristics, past party voting, and family predispositions.

The authors took two important theoretical steps. They argued that voters compare the candidates. Then they made candidate evaluations virtually synonymous with the vote: It is the only variable having a direct impact. Page and Jones consequently continued the trend toward characterizing the voter’s choice in terms of candidates rather than par...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Table of Contents
• Preface
• Part I: Viewing Political Television: Models and Theories
• Part II: Psychological Processing of Issues, Images, and Form
• Part III: Differential Processing of Positive and Negative Advertising
• Part IV: The Psychological Contexts of Processing
• Author Index
• Subject Index