Suppose that you were hired as a consultant to B College to examine the factors which affect a student’s academic performance at B College. You conducted a survey asking students about their GPA, gender (male, female), race (whites, non-whites), major (science, humanities, social sciences, etc.), class (freshman, sophomore, junior, and seniors), co-curricular activity (participation in athletics, artistic performances, or campus organization), class attendance, hours spent studying and completing assignments per week, and parental education.
a. Write an econometric model which allows you to estimate the factors that determine students’ academic performance at BCollege. Be sure to specify your binary variables for race, gender, major, co-curricular activity, and class and interpret the coefficient on each of the binary variables.
b. Suppose that your dependent variable, GPA, is measured with error (students’ reported GPA may differ from their actual GPA*). Discuss how a measurement error in GPA will affect your results.
c. Suppose that one of your independent variables, hours spent on studying, is measured with errors (students’ reported study hours may differ from their actual study hours). Discuss how a measurement error in study hours will affect your results. Do you think that the measurement error is likely to be correlated with the actual study hours or reported study hours?
d. Suppose that parental education is a proxy variable for unobserved family support structure. Discuss the conditions under which you can still obtain biased but consistent estimates of the determinants of students’ academic performance at B College.