Improving College Performance and Retention the Easy Way: Unpacking the ACT Exam

ACT scores are increasingly popular criteria for college acceptance, and are often used—mainly through the ACT’s own “College and Career Readiness” reports—as gauges of college readiness. But how well do they really reflect student achievement in high school, and how well do they predict success in college? This NBER working paper separates the four subject tests that comprise the ACT composite score—mathematics, English, reading, and science. It finds that higher scores on the mathematics and English exams are correlated with higher high school and college GPAs and with lower college dropout rates, while reading and science scores provide virtually no predictive power regarding student success. (These findings are robust, even when controlling for student demographics, college majors, and the selectivity of the colleges that students attend.) Going further, the authors conclude that, in considering students’ composite ACT scores rather than their math and English scores, colleges are “undermatching” some students, meaning that selective colleges are not admitting the highest-performing students possible. The authors determine that, if colleges looked only at math and English scores, as many as 55 percent of students would attend different colleges without significantly disrupting the racial and gender distribution of students in those schools, and top colleges could reduce their dropout rates by 5 to 7 percent. This analysis is important; but look quizzically at its conclusions. Instead of explaining why the other subjects may be poor predictors of college success (science is deprioritized on college campuses and reading itself is a prerequisite for English), it recommends that only math and English composite scores be used in college admissions. A sad curricular narrowing that would be, indeed.

Eric P. Bettinger, Brent J. Evans, and Devin G. Pope, “Improving College Performance and Retention the Easy Way: Unpacking the ACT Exam,” (Cambridge, MA: National Bureau of Economic Research, June 2011).