9.2 Using the data in Table 7.2 from the 1992 United States presidential election, logistic regression was used to model voting for Clinton (i.e., yes rather than no) from the voter’s political view (liberal, moderate, or conservative). Using dummy coding for political view

(X) where conservative is the reference category, X1

= 1 if liberal, and X2

= 1 if moderate, the following parameter estimates were obtained: ˆ ˆ ˆ α β = −0 9. , 2 1  . = = 86, 1 0. . 3 1 2 andβ

a. Use a computer software package to fit the logistic regression model to these data and obtain these parameter estimates.

b. What does each of the parameter values represent? Interpret these values fully and substantively.

c. Change the reference category to liberal and report if and how the parameter estimates changed. What does each of the parameter values represent using liberal as the reference category?

d. Based on these parameter estimates, what are the predicted log odds, odds, and probability of voting for Clinton if one is moderate?