Eco311, Second Midterm Exam, Fall 2016.

Prof. Bill Even

-1-

Your Name (Please print) ________________________________________

Your section (Please circle) 10:00 11:30

Your row (Please circle) 1 2 3 4 5

Directions

Place your answers to all questions in the space provided. Clearly circle your answer to each question

so that it can easily be distinguished from your work. Round all numerical answers to the nearest 100th

(e.g. 1.23) unless specifically told otherwise. Each question is worth 4 points unless otherwise

indicated.

The formula sheet and tables with the standard normal CDF and F-distribution are attached to the end

of the exam.

Eco311, Second Midterm Exam, Fall 2016.

Prof. Bill Even

-2-

The table below was generated from a regression of standardized math scores measured in the 12th grade

as a function of the following characteristics:

motheduc, fatheduc: years of education for the student’s parents

cathhs: a dummy variable that equals one if the student attended a Catholic high school

lfaminc: log of family income

female: a dummy variable that equals one if the student is female.

black: a dummy variable that equals one if the student is black.

1. Construct a 99% confidence interval for the effect of father’s education on the math score.

+.𝟖𝟖𝟕±2.57*.055 =(.746, 1.028)

2. Suppose that school funding improves math scores. Adding a control for school funding to the

regressions would cause the estimated coefficient on the black dummy variable to

a. Increase

b. Decrease

c. Increase only if black students are in schools with lower funding than white students

d. Decrease only if black students are in schools with lower funding than white students

3. Suppose that better school funding increases student test scores. Adding a control for school

funding to the regressions

a. Will definitely not decrease R2, but could decrease adjusted R2

b. Will definitely not decrease R2 or adjusted R2

c. Will definitely not increase R2 or adjusted R2

d. Will definitely not decrease adjusted R2, but could decrease R2

_cons 15.61951 1.32827 11.76 0.000 13.01572 18.22329

black -5.23345 .3896062 -13.43 0.000 -5.997189 -4.469712

female -1.069473 .1959767 -5.46 0.000 -1.453643 -.6853035

lfaminc 1.457989 .143305 10.17 0.000 1.177071 1.738908

cathhs 1.5192 .412244 3.69 0.000 .7110846 2.327315

fatheduc .8872497 .0553613 16.03 0.000 .7787258 .9957735

motheduc .7574657 .0613026 12.36 0.000 .6372952 .8776362

math12 Coef. Std. Err. t P>|t| [95% Conf. Interval]

Total 664709.043 7,429 89.4749015 Root MSE = 8.4235

Adj R-squared = 0.2070

Residual 526696.571 7,423 70.9546775 R-squared = 0.2076

Model 138012.472 6 23002.0787 Prob > F = 0.0000

F(6, 7423) = 324.18

Source SS df MS Number of obs = 7,430

Eco311, Second Midterm Exam, Fall 2016.

Prof. Bill Even

-3-

4. Suppose that better school funding increases student test scores. Also, suppose that family

income is positively correlated with mother and father’s education, but is uncorrelated with the

student’s sex (female). Adding a control for student funding to the regressions will

a. Definitely increase the standard errors for the estimated coefficients on female,

motheduc, and fatheduc

b. Could increase or decrease the standard errors for any estimated coefficients, but is more

likely to increase the standard error for the coefficient on female than that on motheduc

or fatheduc.

c. Could increase or decrease the standard errors for any estimated coefficients, but is

more likely to increase the standard error for the coefficient on motheduc and fatheduc

than that on female.

d. Definitely decrease the standard errors for the estimated coefficients on female,

motheduc, and fatheduc

Other things being the same, indicate whether the standard error of the estimated coefficient on mother’s

education would be larger (L), smaller (S), or unaffected (U) if

5. Mother’s education (not the coefficient on mother’s education) had increased variance __S___

6. Mother’s education was more highly correlated with father’s education __L_____

7. The variance of the regression residual was greater __L___