Statistical models for college admission and enrollment: A case study for a selective liberal arts college

Abstract

In making admission decisions, selective colleges make choices that allow them to improve the quality of the college. Desired student attributes include leadership, noteworthy achievements and talents, and cultural and geographical diversity, as well as high academic performance. Since few applicants dominate in all these attributes, a delicate balancing act must be forged.

In this study we construct probability models for college admissions and student enrollment. We have chosen a selective liberal arts college as a case study over a four-year period. In the first stage we estimate the probability of acceptance to the college for individual applicants. The logit model includes a number of academic factors as well as nonacademic components. In the second stage we estimate the probability of enrollment for an accepted student.

The first stage results show that the actual outcome of the admissions decision does match the broader institutional goals of the college. However, in the second stage, the discrepancy between the acceptance and enrollment decisions frustrates the efforts of the institution to achieve its diversity goals and highlights the difficulty that admission officers have in targeting their efforts toward special groups of applicants.