Continuing Problem 8 on the 2009 golfer data in the file P10_08.xlsx, the simple linear regressions for Earnings per Round were perhaps not as good as you expected. Explore several multiple regressions for Earnings per Round, using the variables in columns I–M and R. Proceed as follows.
a. Create a table of correlations for these variables.
b. Run a regression of Earnings per Round versus the most highly correlated variable (positive or negative) with Earnings per Round. Then run a second regression with the two most highly correlated variables with Earnings per Round. Then run a third with the three most highly correlated, and so on until all six explanatory variables are in the equation.
c. Comment on the changes you see from one equation to the next. Does the coefficient of a variable entered earlier change as you enter more variables? How much better do the equations get, in terms of standard error of estimate and R2, as you enter more variables? Does adjusted R2 ever indicate that an equation is worse than the one before it?
d. The bottom line is whether these variables, as a whole, do a very good job of predicting Earnings per Round. Would you say they do? Why or why not?