Uncertainty and Campaigns

• Published date: 01 January 2015
• DOI: 10.1177/1532673X14535075
• Author: David A. M. Peterson
• Date: 01 January 2015
• Subject Matter: Articles

American Politics Research

2015, Vol. 43(1) 109 –143

© The Author(s) 2014

Reprints and permissions:

sagepub.com/journalsPermissions.nav

DOI: 10.1177/1532673X14535075

apr.sagepub.com

Article

Uncertainty and

Campaigns: The

Psychological Mechanism

Behind Campaign-

Induced Priming

David A. M. Peterson1

Abstract

Campaigns change how some people vote. How campaigns have this effect

is less well understood. The prevailing view is that these effects occur

by changing the content of voters’ attitudes such as partisanship or issue

positions (persuasion) and by changing the weights voters applied to these

determinants of vote choice (priming). Recent research has challenged

this view and suggests that the support for these priming and persuasion

effects is overstated. Unfortunately, no research directly specifies and tests

the specific psychological mechanism responsible for campaign priming. In

this article, I draw on the differences in the forms of attitude strength and

demonstrate that changes in citizens’ uncertainty are responsible for these

effects. The results suggest that persuasion and changes in uncertainty (but

not ambivalence or importance) are responsible for the changes in voters’

decisions during the campaign. Substantively, the largest effects occur

because of changes in the uncertainty voters have about the nature of the

candidates’ character traits.

Keywords

campaign, priming, attitude strength, uncertainty

1Iowa State University, Ames, USA

Corresponding Author:

David A. M. Peterson, Department of Political Science, Iowa State University, 541 Ross Hall,

Ames, IA 50011, USA.

Email: daveamp@iastate.edu

535075APRXXX10.1177/1532673X14535075American Politics ResearchPeterson

research-article2014

110 American Politics Research 43(1)

Campaigns alter some voters’ decisions (Campbell, 2000; Hillygus &

Jackman, 2003; Holbrook, 1996; Shaw, 1999; Wlezien & Erikson, 2008). If

the campaign did not exist, many voters would make a different choice from

the one they do on Election Day. The long-standing view of presidential elec-

tion campaigns is that they change voters’ minds by persuading and priming

relevant attitudes (Bartels, 2006; Johnston, Blais, Brady, & Crete, 1992;

Johnston, Hagen, & Jamieson, 2004). Priming, in this context is when a cam-

paign focuses on a particular topic and voters’ attitudes about the topic

become more strongly linked to their vote choice, a process Riker (1990)

called heresthetic change. The traditional test of priming is simple: Calculate

the weights applied to an attitude in a vote choice model at one point in time,

and then recalculate the weights applied by the same individuals at a second

moment in time. If the weights increased, the attitude had been primed. For

the past 25 years, the evidence for this seemed incontrovertible.

Recently, Lenz (2009, 2013) has challenged this conventional view of

campaigns by noting that the existing evidence for priming effects are obser-

vationally equivalent with a very different explanation. Instead of the atten-

tion to a topic strengthening the link between that topic and vote choice,

voters may respond to the increased salience by changing their attitudes on

the topic so that they are consistent with their vote choice. Instead of becom-

ing more liberal and, thus, being more likely to vote for the Democratic can-

didate, they become more liberal because they are supporting the Democrat.

Lenz’s results suggest that campaigns do not lead voters to their choices and

that voters do not use elections as opportunities to lead policy.

These are two very different understandings of how campaigns and voters

interact. Under priming theory, voters have preexisting and relatively fixed

attitudes about the topics of the campaign. The information environment

serves to help them decide the weights applied to these considerations. Lenz’s

depiction of voters is less flattering. They form an initial vote choice and shift

their attitude to match this choice. Candidates cannot move voters far and

voters do not respond much to the information they gain. There may be sys-

tematic movement in the underlying attitudes, but the campaigns have mini-

mal influence over voters’ choices.

Lenz’s view of voters’ responsiveness to campaigns may be overly pessi-

mistic, however. Both Lenz’s approach and the priming literature share two

features that may limit their ability to uncover these campaign effects. First,

they both look at single issues in isolation rather than combine a battery of

issue questions into a broader measure, a more common and superior

approach in the issue voting literature (Alvarez, 1997; Alvarez & Nagler,

1998; Ansolabehere, Rodden, & Snyder, 2008; Blais, Turgeon, Gidengil,

Nevitte, & Nadeau, 2004; Treier & Hillygus, 2009). Second, and more impor-

tant, neither Lenz nor the priming studies measure and test a mechanism for

Peterson 111

these changes in the weights applied. They implicitly assume that all voters

should have the same shift in weights and do not measure the psychological

mechanism responsible.1 In this article, I test if changes in a voter’s uncer-

tainty, ambivalence, and importance are responsible for the changing weights

applied to attitudes during a campaign. The results suggest that campaigns

influence voters by helping them become more certain about the candidates.

The article proceeds as follows. In the next section, I develop hypotheses

about how campaigns change the predicted probability of voter’s choice. I

focus on the role of shifts in the content of citizens’ attitudes and the changing

strength of these attitudes. I then move to a discussion of the data, the

Presidential Campaign Impact on Voters: 1976 Panel developed by Patterson

(1976).2 In particular, I address the nature of this panel survey and why it is

the best available data for testing this question. The next section of the article

presents the method for analyzing these data, a nonlinear growth curve. I then

present the results of the statistical models and conclude with what these

results mean for the study of campaigns and how these findings need to be

extended.

Campaigns and Vote Choice: The Problems With

Priming

A voter’s choice in November is a result of three things. First, a voter enters

the campaign with predispositions about politics that provide a starting place

for his or her vote. Their partisanship, issue positions, and other precursors to

vote choice lead them to a baseline decision. For most voters, this baseline is

the way they vote in November (Finkel, 1993; Markus, 1982; Patterson,

1980). The other two portions of a voter’s choice are due to the campaign.

Bartels (2006) and Lenz (2013) suggest that campaigns change a voter’s

decision from this baseline through two processes: persuasion and priming.

Persuasion, or opinion change, is the change in the content of an individual’s

antecedents of vote choice (e.g., issue positions). Priming occurs when the

attitude does not change but the weight that a voter applies to the attitude

changes.

There is ample work believed to document this priming effect in cam-

paigns (Bartels, 2006; Carsey, 2001; Gelman & King, 1993; Johnston et al.,

1992; Johnston et al., 2004). These works share a common design for testing

for this priming effect. They rely on time as the central measure of the cam-

paign, comparing the coefficients from one moment in time to the coeffi-

cients at a later time. If the coefficients increased, the attitude is a stronger

predictor of vote choice and the campaign primed the voter.