That’s pretty much the bottom line in Nate Silver’s regression analysis of the impact of 23 demographic factors on partisan voting in the 2008 election. His object was to see how much of a difference union membership makes to the likelihood of voting Democratic. The answer is: in the same ballpark as evangelicals and weekly worship attenders (God gap) were likely to vote Republican.

Of course, the relative impact of these demographic preferences varied, because the size of each group varies About 10 percent of 2008 voters were union members. Roughly a quarter were evangelicals. And nearly a half were weekly attenders. Silver calculates that the union vote boosted Obama’s (and the Democratic congressional) total by 1.2 percent. Do the arithmetic and that means a three percent boost from evangelicals and a six percent boost from weekly attenders for the GOP. Look for both of those numbers to increase in 2012.

uniondem1.png

Categories: Beliefs

Mark Silk

Mark Silk

Mark Silk is Professor of Religion in Public Life at Trinity College and director of the college's Leonard E. Greenberg Center for the Study of Religion in Public Life. He is a Contributing Editor of the Religion News Service

4 Comments

  1. “Silver calculates that the union vote boosted Obama’s (and the Democratic congressional) total by 1.2 percent. Do the arithmetic and that means a three percent boost from evangelicals and a six percent boost from weekly attenders for the GOP.”
    I’m not a math wiz but considering that Obama won the election by a substantial margin, the relative impact (or boost) that evangelicals and weekly attenders gave to the GOP (McCain) was likely even higher than your estimate. After all, the GOP lost and got fewer votes (though that’s not always the case: Florida 2000, boo-yah!)

  2. That depends on the distribution of the independent variables and on what is presented in the graph, which is unmarked in Silver’s article. Given that it is a logistic regression, it is most likely that Silver is presenting the exponentiated logits for the multiple logistic regression model. Given the relatively low rates of union membership, the overall impact for Obama may be lower than the overall impact of Sectarian Protestants for McCain–even without considering church attendance. That is because there are more sectarians than there are union members, and the predictive model multiplies the logit by the proportion to contribute to the overall prediction.

Leave a Reply

Your email address will not be published. Required fields are marked *

Comments with many links may be automatically held for moderation.