https://bit.ly/4exvnML
https://bit.ly/4cvKCnT
https://bit.ly/3VQtuUl
https://bit.ly/3VMa5DY
https://bit.ly/3VOMUJ2
https://bit.ly/4crbjd5
https://bit.ly/4ewObf6
https://bit.ly/4ewOcja
https://bit.ly/4ewOcQc
http://www.corrections.com/events/12439
https://start.me/w/XXzlG9

At any rate my statistical point is, I wonder to what extent the trappings of subculture are impacting the data separate from race? At dating age, I’d honestly likely have passed on a message from black guy standing in front of a gangsta rap poster, but replied to the same black guy standing in front of a Jimi Hendrix or Bob Marley or RATM poster. If it were possible to control for signs of subculture in clothes and in photo background, you might see a substantial amount of the race differences disappear… or you might not. More realistically than sorting through a zillion photos and giving a subjective subcultural categorization, maybe if you have searchable fields for what music people like, you could use the categories record companies or Amazon.com uses to classify music, and use those categories as an approximate subcultural control.

3. Controlling for physical attractiveness

You say, “We were careful to preselect our data pool so that physical attractiveness (as measured by our site picture-rating utility) was roughly even across all the race/gender slices. For guys, we did likewise with height.”

I’d be interested in the details of how you did this. Attractiveness is such a massive influence that even small differences could throw the numbers out of whack. And there could be weird effects to account for. For example, if the mean of one race centers around a two and the mean of another around a four (plausible because diets vary so much by culture, and nutrients impact affects symmetry, while sugars and calories impact skin and weight), then possibly reviewers start thinking “this photo is a 4/5 for an X girl”, where X is a lower-averaging race, but if they were thinking more in overall terms maybe they’d be saying the photo is a 3/5. And then you go in using the number as a four and not a three, when really the four came about because the viewer was somehow (subconsciously?) adjusting for race to some degree. And then it looks like men should be replying to race X girls at a higher rate than they are, when really it is the attractiveness ratings that throw things off.