You don't multiply percentages like that to calculate voters with multiple characteristics that would suggest a voting percentage. That's like saying, I'm a male and Republicans only get 50% of the male vote and I'm Catholic and Republicans only get 50% of the Catholic vote, so being a male Catholic, the Republicans should only have a 25% chance of getting my vote.
As it relates to the question, we'll probably never get reliable exit polling on how gay blacks vote and we'd have to try and answer which characteristic they would be more likely to identify as first, among other things to estimate a percentage.
Yeah, I should've put a bit more thought into that post.
