Tuesday, March 26, 2019

Data: What are the best predictors of voting for Trump?

It's always exciting for nerds like me when new GSS data arrives. 2018 is now available. They asked respondents whether they voted for Clinton or Trump. I used logistic regression to determine what predicts voting for Trump (sample size = 904):

Voted for Trump 
Male  .57
Southern region  .33
Black  -3.78
Other race  -1.30
Education  -.11
Income  .01
Church attendance  .16

The coefficients are not standardized, so they can only tell you the direction of the relationships, not the magnitudes. I included age and IQ, but neither one was significantly related to vote choice.  Keep in mind that the relationships are net of the influences of the other factors included in the model, so, for example, age might predict voting for Trump because older people are whiter, wealthier, less educated and more religious, not because they grew up in an earlier era.

So here are the characteristics of people voting for Trump: male, Southern, white, less educated, higher income, with more church attendance. No surprises there. And what mattered the most in order from most to least predictive: 1) white, 2) church attendance, 3) male, 4) education, 5) income, and 6) southern residence.  Trump owes the election to religious, white males, like myself.  You're welcome.

I didn't have data on the impact of Russian colluders.

Interpreting Your Genetics Summit


  1. Anonymous9:31 AM

    Being married did not predict support for Trump?

  2. I didn't look, but I'm sure it does.


Who's smarter--Trump or Hillary voters?

This new article from Medium.com concludes that Trump voters are 3-5 points dumber than Clinton voters. Wrong. The General Social Survey ...