Categorical Lab #5 Assignment
Please use the "turnout" dataset within the Zelig package, which details a random subset of 2000 individuals (out of 15,837 surveyed) during the 1992 presidential election. There are 5 variables in these data: race (whether the participant was "white" or nonwhite ["others"]), age (how old the participant was at the time), educate (the participant's years of education), income (a categorical measure of the participant's level of income), and vote (whether the participant voted or not [1 = voted, 0 = did not vote]).
To access this dataset, first call the Zelig package [i.e., library(Zelig)], then use data(turnout) to "attach" the dataset to R. You may then use the turnout dataset in R.
1. Run a logistic regression, with vote as the dependent variable and educate as the independent variable. Please report the beta estimate of the effect of educate on vote, and provide a meaningful interpretation of this estimate.
2. Provide a 95% profile likelihood ratio confidence interval and a 95% Wald confidence interval of the beta estimate in Question 1. Translate the intervals into odds and provide a meaningful interpretation of each confidence interval. According to each confidence interval, is the estimate significant?
3. Create a probability plot for the logistic regression in Question 1. Provide this plot. Calculate the probability for the following questions: What is the probability of voting at a high school education (12 years)? At a college education (16 years)? At the mean level of education?
4. What is the median effective level of education for this model? In other words, what is the value of education at which the probability of voting is 50%? What significance does this point have for the rate of change of the probability?