Even though LGB Californians may be more likely to live in coastal communities, coastal residence would NOT statistically account for LGB Californians’ opposition to Prop 8, because these are two separate sources of influence. LGB Californians in the central valleys were certainly more likely to vote against Prop 8 than their heterosexual counterparts, so a statistical model would identify two separate causes of Prop 8 opposition: Coastal residence and LGB status. These are non-redundant signals.

If it is established that accounting for religiosity eliminates racial differences in Prop 8 support (but not vice versa), religiosity can be understood as a more “proximal cause” of Prop 8 support. I use the word “cause” here loosely, in that there could be something in people (Black and White alike) that attracts them to religion and leads them to be anti-gay. If religion is a better marker of this “anti-gay proclivity” than race, it could both be true that Blacks have more of this anti-gay proclivity than Whites, but that religion is a more efficient marker of this proclivity. I think this is the point Timothy was making; establishing what “proximally explains” an effect is different from understanding its root cause.

To address the yarmulke example: Imagine there are 10 Jews (all of whom were wearing yarmulkes) and 10 Protestants in a room. 9 of the Jews are liberal, and only 1 of the Protestants is. In this analysis, yarmulke-wearing would not be a predictor of political liberalism after controlling for religion (because it is a redundant signal). The reason that religion accounts for Prop 8 above and beyond race is that it is not a redundant signal.

I don’t think so. Most Christians are anti-gay, its easier to read modern translations of the bible as anti-gay then not. It stands to reason that those who attend church are more likely to be anti-gay.

In other words, if you’re trying to figure out why African Americans voted at higher rates for Proposition 8 than the general population, part of the answer is that they as a group are more religious than the general population–and religious people voted at high rates for Prop. 8.

Let’s see if that logic holds up well in looking at another demographic:

If you’re trying to figure out why gays voted at higher rates AGAINST Proposition 8 than the general population, part of the answer is that they as a group are more likely to live in coastal areas of the state than the general population – and coastal residents voted at high rates against Prop. 8.

Hmmm… not so likely.

Perhaps, Jamie Grant is making a distinction without a difference and is not ignoring the relationship between communities and religion. To suggest that religious attendance is the reason “why” African Americans voted for Prop 8 is much like suggesting that observant Jews vote in a particular way because they wear yarmelkes.