'Responding to sexist attack - US experiment'
(AsPredicted #177943)
Download PDF

---

---

1) Have any data been collected for this study already?
No, no data have been collected for this study yet.

2) What's the main question being asked or hypothesis being tested in this study?
We ask four research questions (likely tackled in four separate articles):
RQ1. How are sexist political attacks perceived by respondents? How are the target and sponsor of these attacks evaluated?
For this RQ, we test the following hypotheses:
H1: Sexist (vs. non-sexist) attacks are more strongly punished by voters 
H1a: Evaluation of the sponsor become more negative after exposure to sexist (vs. non-sexist) attack
H1b: Evaluation of the target become more positive after exposure to sexist (vs. non-sexist) attack
H2: Explicit sexist (vs. Implicit sexist) attacks are more strongly punished by voters
H3: Sexist attacks by a Democrat (vs. Sexist attacks by a Republican) are more strongly punished by voters
RQ2. When female candidates are attacked with sexist remarks, to what extent are their responses to the attacks (e.g., counterattacking with sarcasm) "successful" to shape their image in the eye of respondents?
For this RQ, we test the following hypotheses:
H1: Not responding to sexist attack is punished by voters
H1a: Evaluation of the target become more negative after exposure to a non-response (vs. any other response)
H1b: Not responding (vs. any other response) to an explicit (vs. Implicit) sexist attack is punished by voters
H1c: Not responding (vs. any other response) to a sexist attack by a Democrat (vs. Sexist attack by a Republican) is punished by voters
H2: Not confronting sexism is punished by voters
H2a: Evaluation of the target become more negative after exposure to a non-confronting sexism (vs. Confronting sexism) response
H2b: Not confronting sexism (vs. Confronting sexism) to an explicit (vs. Implicit) sexist attack is punished by voters
H2c: Not confronting sexism (vs. Confronting sexism) to a sexist attack by a Democrat (vs. Sexist attack by a Republican) is punished by voters
To test for the effectiveness of responding to sexist attacks (confronting the sexism), we run an exploratory analysis on the 6 response conditions (educational; argumentative – civil; argumentative – uncivil; empathetic; humorous – irony; humorous – sarcasm). Here, we are also examining the moderating role of the type of sexist attack (explicit or implicit) and the partisanship of the woman candidate (Democrat or Republican).
RQ3. What are the systemic consequences of sexist political attacks, in terms of radical partisanship (support for violence, partisan Schadenfreude, moral disengagement, social distance)?
RQ4. What are the systemic consequences of sexist political attacks, in terms of sexist attitudes (hostile, benevolent, modern and political sexist beliefs)?

3) Describe the key dependent variable(s) specifying how they will be measured.
There are four sets of dependent variables, corresponding to different experimental phases (see answer to question 4).
In the first set of dependent variables (DV set 1), asked after the first experimental component, respondents are asked to evaluate how much they like the target and the attacker (0-100 slider), and whether they would vote for the two candidates (0-10 slider).
In the second set of dependent variables (DV set 2), asked after the second experimental component, respondents are asked again to evaluate how much they like the target and the attacker (0-100 slider), whether they would vote for the two candidates (0-10 slider), and whether they agree that a series of four personal traits (complainer, irritating, whiny, overreactive) describe the target well (0-10 slider).
Additionally, all respondents are asked, after the second set, to rank all response options uttered by the target after the sexist attack (second experiment), from the best to the worst (DV set 3).
Respondents are finally asked a series of questions to measure radical partisanship and sexist attitudes (DV set 4). More specifically:
(i) Partisan schadenfreude (0-10 slider) (inspired by materials in Webster at al., 2023);
(ii) Moral disengagement (0-10 slider) (based on Kalmoe & Mason, 2022);
(iii) Social distance (0-10 slider) (based on Bogardus, 1933);
(iv) support for partisan violence (0-10 slider) (inspired by materials in Westwood et al., 2022);
(v) to what extent they agree or disagree (0-10 slider) with a battery of statements measuring hostile sexist attitudes (based on Glick and Fiske 2001; short version of scale based on Cassese and Holman 2019);
(vi) to what extent they agree or disagree (0-10 slider) with a battery of statements measuring benevolent sexist attitudes (based on Glick and Fiske 2001; short version of scale based on Cassese and Holman 2019);
(vii) to what extent they agree or disagree (0-10 slider) with a battery of statements measuring modern sexist attitudes (based on Ekehammer et.al. 2000);
(viii) to what extent they agree or disagree (0-10 slider) with a battery of statements measuring political sexist attitudes (based on Van der Pas, Aaldering and Steenvoorden 2022).

4) How many and which conditions will participants be assigned to?
The experimental design is set up in two phases.
In phase 1 (exposure to the attack), respondents are randomly assigned to one of four experimental conditions, presented as a newspaper article describing a debate between two candidates (first half of the article):
- (A1). Man attacks man via non-sexist attack
- (A2). Man attacks woman via non-sexist attack
- (A3). Man attacks woman via implicit sexist attack
- (A4). Man attacks woman via explicit sexist attack
Each of the four conditions exists in two versions, one where the attacker is a Democrat and the target a Republican (a), and one whether the attacker is a Republican and the target is a Democrat (b). This creates 2x4=8 different treatments in total for Phase 1.
After exposure to the attack, respondents are asked a series of questions to assess how they perceive the two candidates (DV set 1).
In phase 2 (response to sexist attacks), respondents who were exposed to a sexist attack in Phase 1 (conditions A3 and A4) are randomly assigned to one of eight experimental conditions, presented as a newspaper article describing a debate between two candidates (second half of the article):
- (R1). No answer
- (R2). Target does not engage with sexism in the attack received
- (R3). Educational response
- (R4). Argumentative response (civil)
- (R5). Argumentative response (uncivil)
- (R6). Empathetic response
- (R7). Humorous response (irony)
- (R8). Humorous response (aggressive sarcasm)
As the candidates in the exchange are the same as in Phase 1 (continuation of the debate), these responses exist in four versions, based on whether the attacker is a Democrat and the target a Republican (a), or whether the attacker is a Republican and the target is a Democrat (b); and based on whether the attack was implicitly sexist (a) or explicitly sexist (b). This creates 2x2x8=32 different treatments in total for Phase 2.
After exposure to the response in Phase 2, respondents are asked a series of questions to assess how they perceive the two candidates (DV set 2).
All respondents, including those who were not included in Phase 2, are then asked to rank all response options (DV set 3), as well as a series of questions to measure their radical partisanship and sexist attitudes (DV set 4).

5) Specify exactly which analyses you will conduct to examine the main question/hypothesis.
OLS regressions (with and without controls), regressing the DV's on the assignment to experimental conditions.

6) Describe exactly how outliers will be defined and handled, and your precise rule(s) for excluding observations.
Participants are excluded if they fail to correctly answer the attention check question. Robustness checks will be run that exclude straight-liners (SD=0 on batteries with reversed statements) and 1% fastest and slowest respondents.

7) How many observations will be collected or what will determine sample size?
No need to justify decision, but be precise about exactly how the number will be determined.
We plan to survey 8,000 respondents. We used GPower 3.1 for a-priory power analysis and calculated that we need 176 participants per experimental group to detect small direct effects (d=0.3) within a t-test comparing means of independent groups, with alpha=0.05 (two-sided) and 80% power (balanced setting). Because we have up to 36 experimental groups ((sexist treatments: 8 possible answers to the sexist attack * 2 types of attack (explicit, implicit sexism) and * 2 partisanship of the attacker (Democrat, Republican)) + (non-sexist treatments: 2 partisanship of the attacker (Democrat, Republican) * gender of the attacker (man, woman) = 36), we would need a minimum of 176 * 36 = 6,336 respondents to capture these effects. We have decided to oversample towards a target of 8,000 respondents to also account for possible excluded observations (failed attention check, straight-liners, slow and fast respondents).

8) Anything else you would like to pre-register?
(e.g., secondary analyses, variables collected for exploratory purposes, unusual analyses planned?)
This pre-registration will likely result in four different articles:
Article 1 will investigate the effect of exposure to the (sexist) attack on candidate evaluation (RQ1). Data collected in the USA will be compared to existing data already gathered for Germany (see Aspredicted 100918; https://aspredicted.org/267_2F2)
Article 2 will investigate the effect of the different responses on candidate evaluation (RQ2)
Article 3 will investigate the systemic effect of exposure to (sexist) attacks, as well as possible responses, on respondents' radical partisanship (RQ3).
Article 4 will investigate the systemic effect of exposure to (sexist) attacks, as well as possible responses, on respondents' sexist beliefs (RQ4).
Additionally, in an exploratory analysis we will also study the moderated effects of the (sexist) attacks and the responses, by testing conditionality based on a variety of respondent characteristics (e.g., gender, partisanship, strength of partisan and gender identification).