Conversation

There's a trope that white women fuck dogs, so I asked 5k women their ethnicity, and if they were sexually aroused by dogs. Here's the data (if ppl said they were into it = 1 point; if marked as fav out of similar options = 2; for reference, interest in 'spanking' scored 0.7)

313

Aella

@Aella_Girl

The averaged responses from each category: White women (n=3565): 0.16 Hispanic/latino (n=446): 0.13 East Asian (n=262): 0.1 Asian (other) (n=386): 0.12 Black (African American) (n=137): 0.12 Black (other) (n=67): 0.05 Native American (n=26): 0.08 Pacific Islander (n=24): 0.54

6:03 PM · Jul 6, 2022

Replying to

So it looks like white women are slightly more interested in dog-specific bestiality, but not significantly; the white-ethnicity and dog-interest correlation was r=0.03 Pacific Islander women seem most sexually interested in dogs, but the sample size for this category was low

158

Aella

@Aella_Girl

Jul 6, 2022

(general disclaimer there's a chance I made an error looking into this data, as I'm using python and I'm new to it. I did spot-check tho!)

110

Kareem Carr | Data Scientist

@kareem_carr

Jul 6, 2022

Replying to

@Aella_Girl

I would rescale: 0 for no interest and 1 for anything more. Using your own scale is a bit arbitrary and makes it harder to interpret the average. The evidence here suggests to me these numbers don’t differ at all by race.

Replying to

I'd like to rescale, but haven't yet figured out how to do this in python.

Show replies

Replying to

Thanks for the data! Any reason why you did not include a measure of error here? There are lots of ways it could be done and very few wrong ways to do it (happy to suggest a few if you'd like!). If lack of space consider excluding <100 n groups as data are close to meaningless.

Replying to

8%-16% of white women are into it? You really do know all the fun people.

Replying to

I think the easiest way to test for a significant difference would be to code 1 if sexually aroused by dogs at all, otherwise 0. Group all non-white ethnicities and then run a two-sample t-test. Would you be willing to do this or could you please share the data?

Replying to

What is the data for men?

Replying to

do one on horses

Replying to

Would be interesting to see proportions, e.g.: "White women x/n = y" where x is the number of respondents that said they were into it. would make it possible to do chi-squared test(s), etc