Chapter 8
1. Using the data in GPA3, the following equation was estimated for the fall and second
semester students:
Here, trmgpa is term GPA, crsgpa is a weighted average of overall GPA in courses
taken, cumgpa is GPA prior to the current semester, tothrs is total credit hours prior to
the semester, sat is SAT score, hsperc is graduating percentile in high school class,
female is a gender dummy, and season is a dummy variable equal to unity if the
student’s sport is in season during the fall. The usual and heteroskedasticity-robust
standard errors are reported in parentheses and brackets, respectively.
i. Do the variables crsgpa, cumgpa, and tothrs have the expected estimated effects?
Which of these variables are statistically significant at the 5% level? Does it matter
which standard errors are used?
1. Yes, these three variables have the same results with 2 standard errors
2. With the usual OLS standard error
t-statistics of crsgpa: 5.14, t-statistics of cumgpa: 3.01, t-statistics of tothr: 1.167
at the 5% level with n = 269 => critical value of t is 1.645 => crsgpa and cumgpa
are statistically significant
With the heteroskedasticity-robust standard errors
t-statistics of crsgpa: 5.42, t-statistics of cumgpa: 2.61, t-statistics of tothrs: 1.167
at the 5% level with n = 269 => critical value of t is 1.645 => crsgpa and cumgpa
are statistically significant.
It does not matter which standard errors are used because the results are the
same.
ii. Why does the hypothesis H0: β crsgpa=1 make sense? Test this hypothesis against the
two-sided alternative at the 5% level, using both standard errors. Describe your
conclusions.
H0: Bcrsgpa = 1
with the usual OLS standard error => t-statistics = (0.900-1)/(0.175)= -0.57143
with the heteroskedasticity robust standard error => t-statistics
=(0.900-1)/(0.166)= -0.602
at the 5% level for two-side alternative => critical value of t is -1.960 => cannot
reject the null hypothesis.
iii. Test whether there is an in-season effect on term GPA, using both standard errors.
Does the significance level at which the null can be rejected depend on the standard
error used?
Season is a dummy variable equal to unity if the student’s sport is in season during the
fall => season = 1 and season = 0 is the opposite
With the usual OLS standard error => t-statistics = -0.157/0.098= -1.602
With the heteroskedasticity-robust standard error => t-statistics =-0.157/0.080=-
1.9625
With the heteroskedasticity-robust standard error => it is significant at the
10%, 5% for the one-tailed test and for two-tailed test.
Holding other factors fixed
The term GPA for the fall and second-semester students during the fall is 0.157 less
than that without during the fall.
TEST SIGNIFICANT BẰNG STATA (XEM CẢ CÁI T STATISTICS VÀ
P VALUE) ĐỂ KẾT LUẬN CHUNG
2. Use the data in HPRICE1 to obtain the heteroskedasticity-robust standard errors for
equation (8.17).
i. Discuss any important differences with the usual standard errors.
ii. Repeat part (i) for equation (8.18)
ii. What does this example suggest about heteroskedasticity and the transformation
used for the dependent variable?
It shall be noted that the use of logarithmic transformation reduces the
heteroskedasticity and the usual OLS standard error and heteroskedasticity-
robust standard error across the various coefficients only differ by a marginal
value.
The heteroskedasticity-robust standard error is even higher than the usual
OLS standard error for the coefficients for the regression with log-
transformed dependent variable with the independent ones.
