Attempts to assess institutional performance by means of retention rates, student performance on standardized tests, and other raw outcome measures are seriously flawed because such measures fail to take into account the powerful effect of student inputs. In this study, national longitudinal retention data on 52,898 students attending 365 baccalaureate-granting colleges and universities are used to generate formulas for estimating any institution's expected retention rate based on its students' high school grades, admission test scores, and racial and gender composition. Separate formulas are provided for computing estimated degree completion rates four, six, and nine years after freshmen entry. Institutions that wish to assess their effectiveness in retaining students can compare these estimated rates with actual degree completion rates.
