Statistic HW
Â
1. (20 points)Estimate the following equations and report the results in a table:
(a) GPA as a function of SAT score, total hours, athlete, high school percentile
rank, sex, race, high school size, and school size squared
(b) Add an interaction term between SAT and athlete to regression (a)
(c) Add high school rank to regression (b)
(d) Estimate High school rank as a function of all of the regressors in (b)
2. (15 points)Which model do you prefer and why?(There is not necessarily a correct
answer. As long as you use the criteria that we discussed in class correctly, you
will get credit.)
3. (15 points)Interpret the coecient on the interaction term between athlete and
SAT.
4. (25 points)Calculate a variance in
ation factor for high school rank. What does
a high VIF indicate? Is the estimate cause for concern? (pages 259-260 in text)
Report the necessary regression results.
5. (25 points)Use a new regression and a t-test to test whether sat scores di erently
predict gpa for men and women. Report regression results in a table.
I have done the first question and the work I have done is in the file 1 (a microword file).
The data needed in this HW is in ( a STATAÂ file)
http://www.wikiupload.com/XYSI3WDDRE88MYU
