California and the SAT: A Reanalysis of University of California Admissions Data

As part of the University of California's recent reconsideration of the role of the SAT in admissions, the UC Office of the President published an extensive report, UC and the SAT (2001), which examined the value of SAT I Reasoning Test scores, SAT II Subject Test scores, and high school grades in predicting the grade-point averages of UC freshmen (UCGPA), as well as the role of economic factors in predicting UCGPA. The analyses in UC and the SAT were based primarily on data that had been aggregated across freshmen cohorts (1996 through 1999) and across UC campuses. In the current study, by contrast, data were analyzed within campuses and cohorts and then summarized. While some of our conclusions are similar to those in UC and the SAT, others are not. Like the earlier study, for example, our reanalyses showed that, considered collectively, the SAT II tests required by UC (Writing, Math, and a third test of the applicant's choice) are slightly superior to the SAT I as a predictor of UCGPA. But our reanalyses also revealed considerable variability across campuses and freshman cohorts in the predictive value of high school grades and test scores, which was masked in the earlier analyses. Also, our reanalyses did not support the conclusion in UC and the SAT that SAT II scores are "less sensitive" to socioeconomic factors than SAT I scores, an assertion that was often repeated during the SAT debate that took place in 2001 and 2002.