Test Scores and Self-Selection of Higher Education: College Attendance versus College Completion

As a companion paper to our work on students' application and colleges' admission decisions, we have estimated a joint discrete-continuous utility maximization model of college attendance and college completion. The paper is motivated by the possibility that test scores are poor predictors of who will succeed in college and thus may not promote optimal investment decisions and may indeed unjustly limit the educational opportunities of some youth. We find that: (1) College attendance decisions are strongly commensurate with college completion. Persons who are unlikely to attend college would be very likely to drop out of even their "first-choice" colleges, were they to attend. College human capital investment decisions are strongly mirrored by the likelihood that they will pay off. (2) Contrary to much of the recent criticism of the predictive validity of test scores, we find that their informational content is substantial. After controlling for high school class rank, for example, the probability of dropping out of the first-choice college varies greatly with SAT scores. (3) Individual self-selection, related to both measured and unmeasured attributes, is the dominant determinant of college attendance.