7. As an example of impure serial correlation caused by an incorrect functional form, let’s return to the equation for the percentage of putts made (Pi) as a function of the length of the putt in feet (Li) that we discussed originally in Exercise 3 in Chapter 1. The complete documentation of that equation is (JOHN HARVEY DELETED THE STANDARD ERRORS BECAUSE YOU DO NOT NEED THEM) P^i = 83.6 – 4.1Li (9.32) t = – 10.6 N = 19 adj-R2 = .861 DW = 0.48
Test Equation 9.32 for serial correlation using the Durbin–Watson test at the 5-percent level. (2 pts) YOUR ANSWER HERE:
Why might the linear functional form be inappropriate for this study? Explain your answer. (1 pt) YOUR ANSWER HERE:
If we now reestimate Equation 9.32 using a double-log functional form, we obtain (Bob DELETED THE STANDARD ERRORS BECAUSE YOU DO NOT NEED THEM): lnP^i = 5.50 – 0.92lnLi (9.33) t = – 13.0 N = 19 adj-R2 = .903 DW = 1.22 Test Equation 9.33 for serial correlation using the Durbin–Watson test at the 5-percent level. (1 pt) YOUR ANSWER HERE:
Compare Equations 9.32 and 9.33. Which equation do you prefer? Why? (1 pt) YOUR ANSWER HERE:
Leave a Reply