Do you predict childhood sexual assault will correlate with any fetish?
Resolved
NO
Jun 21
M$5,755 bet
I did a survey where I asked people a gradient question about childhood assault - from no (0) to severe (3). If this question correlates more than 0.12 with any sexual fetish (I tested around 300, depending on how you count subcategories), this question will resolve 'yes.' I'm going to check biological male answers only, which is around n=18,000.
EnopoletusHarding
WINNER
0
AndrewHartman
Out of curiosity, why only men? Are they the only ones you have survey data for, or is this related to a different research question?
0
Aella
@AndrewHartman In general I'm checking both separately, because there seems to be pretty meaningful differences between the sexuality of men and women and I don't want to accidentally conflate a thing. And my sample size for men was higher.
0
AndrewHartman
@Aella Yeah, I agree there seem to be statistically significant differences, at least along certain dimensions. I'd be more confident betting YES on this for women, though I still lean to a YES bet here for men as well.
0
JoyVoid
Ah interesting, thanks for that, I had completely missed those markets have been resolved
0
J
Those past results are very surprising to me. Is this just due to the extremely high bar to find a "significant" correlation in the context of so many multiple comparisons? Is she doing the right multiple comparisons correction? Is this the same reason PGS kinda suck even for things that twin/sibling studies have proven are 80% heritable?
0
Aella
@J I'm not sure if I'm doing the things you describe; I'm basically just looking to see if there's a correlation between the number scales of various questions. I haven't been checking p-values or odds ratios or adjusting based on questions asked for this at all, cause my n=18000 and I'm mostly asking manifold questions before I've checked to see what the answer is.
0
ahalekelly
Check any metric against 300 things and you'll find correlations purely by chance
0
JoyVoid
Yeah, I'm not finding those types of markets particularly interesting because I don't feel like I'd learn more if the answer came out one way or another. I'd be interested if it was like "Do you predict [...] correlate with being involved in the BDSM community" or "Do you predict [...] correlate with being submissive" for instance
0
TimP
Tim P bought M$10 of YES
But the probability of finding correlations purely by chance also depend on the sample size, and this one is pretty big.
0
Aella
@AdrianKelly not if ur sample size is whopping.
0
ahalekelly
Adrian is betting YES at 40%
@Aella Ah yeah you're using an r value threshold not a p value threshold, my bad
0