Recall the Box-Tidwell regression model, Yi 1X1 i1 %%% kXk ik i In Section 12.5.2, I described a procedure for fitting

Recall the Box-Tidwell regression model, Yi ¼ α þ β1Xγ1 i1 þ%%%þ βkXγk ik þ εi In Section 12.5.2, I described a procedure for fitting the Box-Tidwell model that relies on constructed variables. I applied this procedure to data from the Canadian Survey of Labour and Income Dynamics to fit the model log2 wages ¼ α þ β1ageγ1 þ β2educationγI þ β3male þ ε

where male is a dummy regressor coded 1 for men and 0 for women. Fit this model to the data by general nonlinear least squares. Are there any advantages to using general nonlinear least squares in place of Box and Tidwell’s procedure? Any disadvantages? Can the two approaches be combined?