Direct and indirect effects of prejudice: sexism, information, and voting behavior in political campaigns

ABSTRACT

This paper examines the effects of gender-based prejudice on candidate evaluation and voting behavior. It uses a unique experimental design to test for direct effects of sexism on candidate evaluation and voting behavior, as well as indirect effects of sexism on these variables via the information that subjects seek out about women candidates. I find that subjects with higher scores on items measuring modern sexism are less likely to vote for female candidates, less likely to vote “correctly” when their preferences most closely align with a female candidate, and rate female candidates more negatively than their male counterparts. I also find that subjects high in sexism search for less information about women candidates and that less information search also leads to lower feeling thermometer ratings, a lower likelihood of voting for women candidates, and a lower likelihood of casting a “correct” vote for a woman. In sum, sexism has both direct and indirect effects on subjects’ voting behavior.

KEYWORDS:

• Sexism
• prejudice
• candidate evaluation
• voting
• experiments

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1 It is also important to note the recent developments in studies of implicit prejudice using tests such as the Implicit Attitudes Test (IAT; Greenwald, McGhee, and Schwartz 1998) which have found evidence of implicit gender-based prejudice (Rudman, Greenwald, and McGhee 2001; Rudman and Kilianski 2000) have also generally not been applied to studies of voting behavior (though Mo 2015 is obviously an exception).

2 DPTE was developed by Richard Lau and David Redlawsk. The software can be accessed at www.processtracing.org.

3 Following the procedures set out by Baron and Kenny (1986) as well as Sobel (1982).

4 204 subjects chose to vote in the Democratic primary, and 99 chose to vote in the Republican primary.

5 This was a multi-investigator study in which manipulations designed to test several different hypotheses were included in the same experiment. All manipulations were randomly assigned, so candidate race and ethnicity should not systematically affect the results of my study and, indeed, do not. Results of analyses that include race/ethnicity can be found in Ditonto 2019.

6 Pictures were taken from state legislators’ websites in states other than those from which subjects were recruited. All were pre-tested and rated similarly in terms of attractiveness, age, and likability.

7 Available in the appendix.

8 More information on the calculation of correct voting is available in the appendix.

9 Negative binomial regressions were also estimated but were essentially identical to the OLS models.

10 Items include: “Women who demand equality these days are actually seeking special favors,” “Women often miss out on jobs because of discrimination,” “Women who complain about harassment cause more problems than they solve,” “A working mother can establish just as warm and secure a relationship with her children as a mother who does not work,” “It is much better for everyone involved if the man is the achiever outside the home and the woman takes care of the home and family.”