Ta-Nehisi Coates asks:
…I don’t want to scapegoat my brown bothers–my sense is that ethnicity is a really bad filter here–for blacks, whites and Latinos. For instance, is homophobia tied to wealth? Is it tied to education? Is it tied to region? What is the best predictor of homophobia? Is it really race? Or is it something like poverty or even church attendance?
The charmingly named “HOMOSEX” variable in the GSS has large sample sizes, so I decided to look into the various relationships. The question is: “What about sexual relations between two adults of the same sex?” The responses are always wrong, almost always wrong, sometimes not wrong at all and not wrong at all. The numeric code goes from 1 to 4, with the lower codes denoting less tolerance of homosexual relations. Because the GSS is just a snapshot I wouldn’t focus too closely on the proportions as we’re in a time of social change, but the differences between demographics are likely robust (all the trends below are ones I am familiar with from years of polling data). For those of you who don’t like the garish frequency tables, I’ve checked the grayscale option. Finally, the abbreviations and spellings are generated by the GSS application; I don’t have control over them.
No surprises, right? Here are intolerant:tolerant dyads:
But obviously I’m not controlling for other background variables. After all, the average number correct on the vocabulary test for whites was 6.22 vs. 4.79 for whites. And blacks are more likely to be religiously fundamentalist, etc. So the race difference might be accounted for by these variables. But let’s look at some trends broken down by race first.
For clarity, this is # correct on a vocab test on the X axis:
Just by inspection here one might assume that controlling for variables can’t explain away the whole race difference. Smart, educated and very liberal blacks are less tolerant of homosexuals than similar whites. In fact, among downscale sectors there isn’t much of a difference between whites and blacks, the difference shows up among the upscale. There isn’t that much of a difference between fundamentalist blacks and whites. There is a big difference between blacks and whites who consider themselves religious liberals; the former are far less homophobic than black fundamentalists, but note that they’re about as gay friendly as white religious moderates.
All that being said, I play around with a multiple regression model by treating some of these categorical but ordinal variables as existing along a numerical interval. That is, HOMOSEX was a dependent variable on a 1-4 interval. The religious variables and age were powerful predictors of the variation in attitudes toward homosexuality, but race not so much (not even statistically significant). I wanted to post the charts above because I don’t necessarily trust these sorts of slap-dash regressions, but my quick & dirty checks imply that race is a less powerful predictor than religion and age.
Update: A few more charts….
Update II: Plotted vs. socioeconomic index for whites, and then blacks.