Search TESS:


Gender and Reactions to Political Campaigns

Download data and study materials


Principal Investigator(s):

Cindy Kam
Vanderbilt University
Home page:

Sample size: 928
Field period: 08/24/2004 - 08/031/2004


This research design focuses on the impact of aspects of political campaigns on women and men's cognitive engagement in politics. Existing research suggests that women may be more motivated by civic, community-oriented concerns (Epstein 1988; Lane 1961; Tolleson Rinehart 1992); therefore, appeals to citizen duty may encourage more cognitive engagement among women than men. In contrast, a horse race frame may be alienating to women; "From describing campaigns as 'horse races' to calling a clever policy initiative an 'end run,' what politics and athletics have in common is a more subtle, clubby atmosphere of masculinity" (Tolleson Rinehart 1992, 144). Following this logic, depictions of politics that use the horse race metaphor, as well as other sporting analogies, are likely to be more effective in engaging men compared to women. The sex of candidates may also differentially pull men and women into thinking about politics (Burns, Schlozman, and Verba 2001).


Interest in the political race: When the political campaign features two male candidates, men will at the baseline be more engaged in the race than women, as measured by interest in the political race. With one male and one female candidate, female subjects are expected to start with a higher baseline level of interest in the campaign.

Issue placement: With two male candidates, men are expected to place the candidates more accurately in the close election condition compared with the baseline condition, and women are expected to be more accurate in the citizen duty condition compared with the baseline condition.
With one male and one female candidate, the impact of the experimental manipulations may be lower among women, but not among men. Men are expected to start with a lower baseline degree of engagement in the campaign, and they are expected respond to the close election condition with more accurate issue placement of the candidates.

Experimental Manipulations:

The experiment includes three characterizations of a political race: a neutral condition, a citizen duty condition, and a close election condition. Additionally, the experiment manipulates the sex of the candidates. In one condition, the two candidates are male; in the other, one of the two candidates is male and one is female. These experimental manipulations test the possibility that various aspects of political campaigns may differentially pull women and men into thinking about politics. The research design allows us to test for the impact of these three aspects of campaigns: citizen duty, closeness of the election, and sex of the candidates on citizen engagement in a political campaign. The manipulations occur within the context of a "newspaper article" describing a (hypothetical) political race. The article explicitly states the positions of the candidates on three issues: services and spending; health insurance; and gun control.

Key Dependent Variables:

The dependent variables are interest in the political race and placement of the candidates on three issues (all of which are discussed in the stimulus article).

Additional Information:

Because of the limitations of conventional television screen resolution, the "newspaper" article was presented to subjects as lines of text on several screens, rather than as a graphic that resembled a newspaper article.

Summary of Findings:

The experimental manipulations had no statistically discernible impact on interest in the election. Analyses of the sex of the candidate manipulation, the framing manipulations, the combination of the two types of manipulations, and their interactions with respondent sex were inconclusive. Individual-level predispositions (gender, strength of partisan identification, and education) predicted interest in this election. Results for issue placement of the candidates were mixed. There were no effects for placement of the candidates on the standard services and spending item. For placement on health insurance, female subjects placed the candidates more accurately when a female candidate was present, but the framing conditions had no additional effect. Male subjects placed the candidates more accurately when a female candidate was present and in the presence of the close election and citizen duty frames. Results for the gun control issue were inconclusive.


The results were mixed at best. The framing manipulations were subtle and may not have been powerful enough to engage subjects. The effect of sex of the candidate may be contingent upon the level of contestation in the electoral race (Atkeson 2003), and this possibility can be tested in future analyses. The impact of the experimental manipulations may also be contingent not upon sex of the respondent but upon gender norms; two such questions appear in the dataset and can be analyzed in the future. The presentation of the stimulus item (as lines of text) rather than as a "newspaper article" may have constrained the impact of the experimental conditions, and this possibility should be tested in future research.


Kam, Cindy D. 2009. "Gender and Economic Voting, Revisited." Electoral Studies, special issue on The American Voter, Revisited, 28: 615-624.

Copyright © 2014, TESS