Consider the following regression model (Table I) that explains cross-sectional variation of
college gpa COLGPA (cumulative college grade points average) as a function of:
HSIZE: size of high school graduating class, in hundreds,
HSIZE_SQ: the square of HSIZE,
HSPERC: academic percentile in high school graduating class (1% is top, 99% is bottom),
SAT: combined SAT score,
FEMALE: dummy gender variable (which is one for females, and zero for males),
ATHLETE: dummy variable (which is one for student-athletes, and zero for student-nonathletes).