5. Your friend is just finishing a study of attendance at Los Angeles Laker regular-season home basketball games when she hears that you’ve read a chapter on serial correlation and asks your advice. Before running the equation on last season’s data, she “reviewed the literature” by interviewing a number of basketball fans. She found out that fans like to watch winning teams. In addition, she learned that while some fans like to watch games throughout the season, others are most interested in games played late in the season. Her estimated equation (standard errors in parentheses) was:
A^t = 14123 + 20Lt + 2600Pt + 900Wt
(500) (1000) (300)
DW = 0.85 N = 40 adj-R2 = .46
where:
At = the attendance at game t
Lt = the winning percentage (games won divided by games played) of the Lakers before game t
Pt = the winning percentage before game t of the Lakers’ opponent in that game
Wt = a dummy variable equal to 1 if game t was on Friday, Saturday, or Sunday, 0 otherwise
Test for serial correlation using the Durbin–Watson test at the 5-percent level. (2 pts)
YOUR ANSWER HERE:
Make and test appropriate hypotheses about the slope coefficients at the 1-percent level. (4.5 pts)
YOUR ANSWER HERE:
Compare the size and significance of the estimated coefficient of L with that for P. Is this difference surprising? Is L an irrelevant variable? Explain your answer. (1 pt)
YOUR ANSWER HERE:
If serial correlation exists, would you expect it to be pure or impure serial correlation? Why? (1 pt)
YOUR ANSWER HERE:
Leave a Reply