Use the data in ALCO5.XLS and statistical software to run a regression where drinks is explained by a constant term, gpa, male, ofage, cig, pot, intra, and white.

Variable Description drinks Number of alcoholic drinks consumed per week gpa Grade point average 0–4 scale male 1 If the student is male; 0 otherwise ofage 1 If the student is 21 or older; 0 otherwise cig 1 If the student uses tobacco; 0 otherwise pot 1 If the student uses marijuana; 0 otherwise intra 1 If the student participates in intramural athletics; 0 otherwise white 1 If the student is white; 0 otherwise (A) Which variables do not attain their expected signs (B) Which variables are not significant at the 5% critical level?
(C) Plot the squared residuals from the regression against gpa. Does the plot show evidence of heteroskedasticity? Explain.
(D) Do a Park test for heteroskedasticity assuming gpa is the “culprit” variable. Show the five-step procedure.
(E) Perform White’s test for heteroskedasticity. Show the five-step procedure.
(F) Run the original regression using weighted least squares.
Assume the variance of the ui’s is proportional to the square of gpa so that 1/gpa (variance of gpa) is the correct weight. Then apply White’s test for heteroskedasticity to the errors from this weighted regression. Does the test indicate the presence of heteroskedasticity? Show the five-step procedure.
(G) Run the original regression using weighted least squares.
Assume the variance of the ui’s is proportional to gpa so that 1/ gpai (standard deviation of gpa) is the correct weight.
Then apply White’s test for heteroskedasticity to the errors from this weighted regression. Does the test indicate the presence of heteroskedasticity? Show the five-step procedure.
(H) Use the Newey-West technique to obtain heteroskedasticitycorrected standard errors with the original regression. Did heteroskedasticity appear to be biasing the standard errors of the structural parameters very much? Explain.