The second study fails to replicate an earlier finding experimentally, with a larger, more representative sample. That means what was taken as evidence for a causal relationship is now suspect. However, inducing masculinity threat then checking whether support for war, homophobic attitudes, and endorsement of dominance hierarchies increase is hardly the only way one might try to measure a causal relationship between male resentment and right wing populism. It could be that circa 2013 there was a causal relationship, which is responsible for the continuing correlation, but nowadays different things lead to right wing populism. It could also be that the intervention doesn't target male resentment in the right ways, and/or the metric used to assess increased support for right wing populism misses something. I'd call this a lack of evidence, but not quite the sort of lack that makes me think it's evidence for the absence of a causal relationship.
In terms of dating, I don't really care about the causal relationship. I don't want to date someone who just happens to be sexist in addition to not sharing my political views. It's hard to imagine being compatible with someone who thinks I'm inherently inferior. The post seems to imply this is unreasonable, but I'd rather not be married, thanks.
A question, not a criticism: You mention a hostile sexism scale with four components and use it in an analysis of 2024 ANES presidential vote data. I followed your hyperlink to the survey itself, but wasn't quite able to match the questions in the survey to the components you listed. (I consulted the survey PDF at https://electionstudies.org/wp-content/uploads/2025/08/
anes_timeseries_2024_questionnaire_20240808.pdf).
Here are the components listed above and what might be the corresponding questions based on the questionnaire PDF: That women ...
(a) complain about discrimination when they lose in a fair competition (maybe that's MODSEXM-COMDISC: "When women complain about discrimination, how often do they
cause more problems than they solve?," with a five-point Likert on always/never);
(b) seek special favors under the guise of equality (presumably MODSEXM_SPECFAV: "When women demand equality these days, how often are they actually seeking special favors?," with a five-point Likert on always/never);
(c) fail to appreciate fully what men do for them (I couldn't find anything comparable. 😬):
(d) are too easily offended (perhaps GENDRES_INNOCENT: "Many women interpret innocent remarks or acts as being sexist," with a five-point Likert on agree/disagree).
I assume I'm missing something, especially since that last question (GENDRES_INNOCENT) is part of the pre-election questionnaire, while the other two I mention above (MODSEXM_COMDISC and MODSEXM_SPECFAV) are in the post-election survey.
Wow, I am flattered that someone checked into this so thoroughly!
The ANES items I used are:
V241607 PRE: Many women interpret innocent remarks or acts as being sexist.
V241608 PRE: Women seek to gain power by getting control over men.
V242280 POST: When women demand equality these days, how often are they
actually seeking special favors?
V242281 POST: When women complain about discrimination, how often do they
cause more problems than they solve?
You're right that there's a discrepancy here, and the ANES questions aren't quite identical to the "hostile sexism" scale in the linked paper. In particular there is no direct correspondence with c) women fail to fully appreciate what men do for them.
This is an error on my part, I used AI to put this all together quickly and didn't notice the difference in this question between the paper and the ANES items.
These questions do straddle both the pre- and post- election surveys but I didn't think that would make a difference here.
Thanks for this thorough response, Jonathan. It helped me connect the dots. I agree that using pre- and post-survey items aren't a big issue here; when I first wrote, I hadn't realized the ANES interviewed the same respondents for both surveys. As for the disconnect between the hostile sexism scale presented in the article you cite and the ANES questions you used in the analysis, I suppose it means you're not quite using a standard or common model for determining sexist attitudes. It doesn't mean the ANES questions you chose weren't measuring something similar, though. And your results don't diverge from that same article's findings about a correlation between measures of sexism and presidential voter preference in 2024. Nor does the finding affect your larger thesis.
Gunitsky's piece relied in part, through his links, on a different battery of questions for determining sexism: a shortened form of the Male Role Norms Inventory (MRNI-VB). But ultimately, this doesn't really matter, since the question isn't correlation: The correlation appears to exist under both measures, however one might assess the measures or their relative value. The question is, as you say, causation, and it's not clear that there aren't multiple causes, a larger cause, or a correlated cause. This is especially true here because virtually every vote for a particular presidential candidate involves a voter's making peace with areas in which they and the candidate disagree. Moreover, even the areas of agreement may be largely beside the point; someone may be voting for a candidate largely or entirely as a vote against the other candidate(s). I don't see this dynamic analyzed as often.
Excuse me, is this article complete? It doesn't seem to present evidence against a causal relationship, nor evidence suggesting a lack of evidence for a causal relationship.
The studies discussed in the middle section of the article are evidence against a causal relationship. The first severely bounds the size of any causation, the second refutes it experimentally.
The second study fails to replicate an earlier finding experimentally, with a larger, more representative sample. That means what was taken as evidence for a causal relationship is now suspect. However, inducing masculinity threat then checking whether support for war, homophobic attitudes, and endorsement of dominance hierarchies increase is hardly the only way one might try to measure a causal relationship between male resentment and right wing populism. It could be that circa 2013 there was a causal relationship, which is responsible for the continuing correlation, but nowadays different things lead to right wing populism. It could also be that the intervention doesn't target male resentment in the right ways, and/or the metric used to assess increased support for right wing populism misses something. I'd call this a lack of evidence, but not quite the sort of lack that makes me think it's evidence for the absence of a causal relationship.
In terms of dating, I don't really care about the causal relationship. I don't want to date someone who just happens to be sexist in addition to not sharing my political views. It's hard to imagine being compatible with someone who thinks I'm inherently inferior. The post seems to imply this is unreasonable, but I'd rather not be married, thanks.
A question, not a criticism: You mention a hostile sexism scale with four components and use it in an analysis of 2024 ANES presidential vote data. I followed your hyperlink to the survey itself, but wasn't quite able to match the questions in the survey to the components you listed. (I consulted the survey PDF at https://electionstudies.org/wp-content/uploads/2025/08/
anes_timeseries_2024_questionnaire_20240808.pdf).
Here are the components listed above and what might be the corresponding questions based on the questionnaire PDF: That women ...
(a) complain about discrimination when they lose in a fair competition (maybe that's MODSEXM-COMDISC: "When women complain about discrimination, how often do they
cause more problems than they solve?," with a five-point Likert on always/never);
(b) seek special favors under the guise of equality (presumably MODSEXM_SPECFAV: "When women demand equality these days, how often are they actually seeking special favors?," with a five-point Likert on always/never);
(c) fail to appreciate fully what men do for them (I couldn't find anything comparable. 😬):
(d) are too easily offended (perhaps GENDRES_INNOCENT: "Many women interpret innocent remarks or acts as being sexist," with a five-point Likert on agree/disagree).
I assume I'm missing something, especially since that last question (GENDRES_INNOCENT) is part of the pre-election questionnaire, while the other two I mention above (MODSEXM_COMDISC and MODSEXM_SPECFAV) are in the post-election survey.
Wow, I am flattered that someone checked into this so thoroughly!
The ANES items I used are:
V241607 PRE: Many women interpret innocent remarks or acts as being sexist.
V241608 PRE: Women seek to gain power by getting control over men.
V242280 POST: When women demand equality these days, how often are they
actually seeking special favors?
V242281 POST: When women complain about discrimination, how often do they
cause more problems than they solve?
You're right that there's a discrepancy here, and the ANES questions aren't quite identical to the "hostile sexism" scale in the linked paper. In particular there is no direct correspondence with c) women fail to fully appreciate what men do for them.
This is an error on my part, I used AI to put this all together quickly and didn't notice the difference in this question between the paper and the ANES items.
These questions do straddle both the pre- and post- election surveys but I didn't think that would make a difference here.
Here's the full code I used to generate my chart
https://github.com/jstray/anes-2024-sexism/blob/main/anes2024_sexism_crosstabs.py
Thanks for reading.
Thanks for this thorough response, Jonathan. It helped me connect the dots. I agree that using pre- and post-survey items aren't a big issue here; when I first wrote, I hadn't realized the ANES interviewed the same respondents for both surveys. As for the disconnect between the hostile sexism scale presented in the article you cite and the ANES questions you used in the analysis, I suppose it means you're not quite using a standard or common model for determining sexist attitudes. It doesn't mean the ANES questions you chose weren't measuring something similar, though. And your results don't diverge from that same article's findings about a correlation between measures of sexism and presidential voter preference in 2024. Nor does the finding affect your larger thesis.
Gunitsky's piece relied in part, through his links, on a different battery of questions for determining sexism: a shortened form of the Male Role Norms Inventory (MRNI-VB). But ultimately, this doesn't really matter, since the question isn't correlation: The correlation appears to exist under both measures, however one might assess the measures or their relative value. The question is, as you say, causation, and it's not clear that there aren't multiple causes, a larger cause, or a correlated cause. This is especially true here because virtually every vote for a particular presidential candidate involves a voter's making peace with areas in which they and the candidate disagree. Moreover, even the areas of agreement may be largely beside the point; someone may be voting for a candidate largely or entirely as a vote against the other candidate(s). I don't see this dynamic analyzed as often.
Anyway, thanks again for your response.
Excuse me, is this article complete? It doesn't seem to present evidence against a causal relationship, nor evidence suggesting a lack of evidence for a causal relationship.
- Fork
The studies discussed in the middle section of the article are evidence against a causal relationship. The first severely bounds the size of any causation, the second refutes it experimentally.