Statistical Modeling, Causal Inference, and Social Science: Income, religious attendance, and voting

Income, religious attendance, and voting

I can't say I have much of an explanation for this, but it's interesting:

There's lots of talk about religion and income and voting, but people don't always know that interactions are important. Getting back to income, religious attendance, and voting, here for related graphs. We've looked at time trends; the next step is to break things up by religious denomination.

Cyrus

Hi Andy,

I submitted last month an exploratory paper to Bob Erikson on whether and to what extent religiosity mediates "income-based" voting a la Meltzer and Richards. (I think Bob should still have it, or maybe he's slipped it back in my dept mail folder...). Anyway, the idea was to start exploring regional and temporal differences in income/religion interactions by fitting separate models for different regions (I used census regions) for different presidential elections from 1980 to 2004. I used two measures of religiosity: attendance, like you use, and depth of religious belief, measured by resposnses to questions about how important religion is in the respodent's life. The basic results were as follows:

(1) There was considerable regional heterogeneity in the degree to which the religiosity was associated with voters choosing "against" the redistributive preferences that one would expect from a Meltzer-Richards type model;

(2) Since 1980, the dampening effect of religion on income-based voting patterns is only apparent in presidential elections in 1992 and after. For these recent elections, it is reasonable to conclude that a surge in politicized religiosity has dampened the redistributive pressure that would be applied by lower income voters had they voted on their presumed material interests. But it is not evident that religion has been a persistent factor in this regard.

(3) In any case where religiosity did have a significant mediating effect, the effects associated with religious participation and depth of belief tended in the same direction, but the effects associated with participation were usually stronger.

Thought I'd share.

Best,
Cyrus
(writing from Bujumbura, Burundi at the moment)

Andrew

Cyrus,

I'm confused. We don't see a "dampening effect of religion on income-based voting patterns". It's the opposite: we see stronger income-based voting patterns among religious attenders than among non-attenders.

David Weakliem

I can think of two possible explanations:

1. An overwhelming majority of people in the US have at least some commitment to religion. Someone who never attends religious services (and says so in a survey) is making a definite statement about their values, and in most cases the values associated with never attending are ones that would lead you to vote against Bush.

2. Lower income people tend to vote based on material interests--higher income people can "afford" to give more weight to views on other issues (e. g., abortion, gay marriage). So the higher your income, the more religiosity matters (in terms of the graph, the gap between the three lines is larger among people with higher incomes).

I'd incline towards (1), since the relationship between income and who you vote for seems equally strong for frequent and moderate attenders. That is, it's the "nevers" who stand out as exceptional.

Andrew

David W.,

Actually, the proportion of "nevers" is pretty high, and it's been increasing in recent decades. (I can't remember the exact numbers but David Park has them.)