Miron Zuckerman is updating his meta-analysis of the religion and intelligence relationship, and asked me to compute some extra analyses of the OKCupid dataset.
Cognitive ability is independent cause of the usual demographics.
Distribution of cognitive ability by religious stance x seriousness. Error bars = 95% confidence intervals.
Gays more religious than bisexuals? Weird. Female bisexuals less than heterosexual men? Smells of sampling bias, but maybe not.