Another Logistic Regression

Submitted by Professor Shortell on Mon, 12/11/2017 - 18:52

Logistic Regression

Let's test a model to predict voting for Trump in 2016.

First, some recodes:
Vote4Trump<-ifelse(V162034a==2,1,0) #This will be our DV
Gender<-'NA'; Gender[V161342==1]<-1; Gender[V161342==2]<-0
V161310r<-'NA'; V161310r[V161310a>=0]<-0; V161310r[V161310a==1]<-1

Now, the model:
summary(glm(Vote4Trump~Gender+V161310r+V161126,family=binomial))
exp(coef(glm(Vote4Trump~Gender+V161310r+V161126, family=binomial)))

EXERCISE: See if you can improve on the fit of the model (lower AIC) by adding one or more blocks. Interpret the results.