Individual Submission Summary
Share...

Direct link:

Campaign Messages & Voter Bias: Evidence from Text Analysis & Conjoint Surveys

Sun, September 2, 10:00 to 11:30am, Sheraton, Beacon B

Abstract

Research finds that voters, on average, perceive of men as more capable candidates and perceive of minorities as less intelligent, poor leaders (Holman 2011, Tesler 2013). Voters' reliance on racial and gender perceptions results in the attribution of incorrect ideological positions based on candidate demographic identity (Kathlene 1994, Lerman 2016). Nevertheless, the discourse on gender, race, and political candidacy results in a seeming paradox. Analysis of electoral outcomes establishes that these beliefs regarding women and minorities do not present a barrier to holding office (Highton 2004, Lawless 2015).

I seek to resolve this paradox of both evidence of voter preferences or beliefs and findings of equal candidate outcomes. Rather than reject the veracity of either body of literature, I focus on the campaign communication of candidates. Even while voters hold attitudes about demographic groups, they likely perceive of candidate identity as a single piece of information to be analyzed and processed (Ditonto 2014). Thus the effects of voters' attitudes regarding gender and race are attenuated by other information (Mo 2014). Do candidates deliver information to counteract detrimental voter perceptions? This research expands political communication and political psychology research, by adding insights on the influence of candidate race and candidate gender on campaign messaging, and the influence of messaging on voter perceptions.

One dimension of campaign strategy is the extent to which candidate gender and racial identity may result in differing messages. Some scholarship claims that candidates strategically communicate information in order to overcome stereotyping (McGraw 2003). This includes the adoption of masculine messaging strategies (Wagner 2017), and discussing owned-issues (Evans 2014). I theorize that candidates are aware of how voters perceive of race and gender, and the degree to which these perceptions undercut their qualifications. Candidates deal with voter perceptions through their campaign strategy and messages (Dittmar 2015).

To analyze this, I construct a dataset comprised of over 6,000 campaign websites for candidates from the 2008-2016 election cycles. My dataset includes variables regarding candidate race, gender, and district demographics. While previous datasets have utilized campaign websites, this data provides information for beyond two congressional cycles and provides information beyond political affiliation. I utilize a Structural Topic Model (STM) in order to test hypotheses regarding the connection between candidate campaign topic and variables of interest, which include race, gender, and party. Computational STM enables analysis that estimates the effect of changes in these covariates on the probability of documents containing specific topics.

Race and gender—which voters use as heuristics to evaluate competency—affect campaign communication by compelling candidates to follow proactive and reactive strategies. Proactive strategies involve drawing upon issue ownership, endowed by identity or party. Reactive strategies entail preemptively addressing stereotypes. Candidates may utilize either one or both of these strategies depending upon which stereotypes their constituents use to evaluate their competency. I find statistically significant evidence that female candidates proactively select owned topics such as education, health-care, and women's issues. Additionally, female candidates counteract deficits in voter perceptions by discussing their background, endorsements, work ethic, and election. Lastly, I find that minority candidates also use reactive strategies, potentially in response to low voter assessments, by discussing endorsements, background, work ethic, and fundraising.

These topics enable candidates to deliver information about quality and thus minimize voter bias. Communication strategies may explain why evidence of voter race and gender bias occurs despite limited evidence of diminished electoral success. To determine whether voter perceptions of candidates are influenced by these messages, I conducted forced-choice, paired-conjoint experiments. Results from the experiments indicate that when provided with information about demographics, White voters evaluate the competencies and issue capabilities of racial minority candidates negatively. Candidates in racial minority groups on average were associated with perceptions of weak leadership, low rationality and intelligence, and weakness in economic and foreign policy. However, when voters were additionally presented with reactive strategy messaging, racial minority candidates attenuated these negative character ratings. Lastly, including proactive strategy messaging further bolstered perceptions of issue competency in civil rights and crime.

In sum, this work provides cutting-edge analysis of strategic campaign messaging and puts to the test its effectiveness in attenuating voter biases.

Author