A research team is on the cusp of releasing a new study that may contribute to the discussion about whether standardized test scores should have a place in the college admissions process.
An abstract for the working paper, published on the National Bureau of Economic Research website, states the researchers analyzed the relationship between students’ standardized test scores, their grades in high school, and grades in college.
“Standardized test scores predict academic outcomes with a normalized slope four times greater than that from high school GPA, all conditional on students’ race, gender, and socioeconomic status,” the researchers wrote.
Some universities recently began requiring SAT or ACT scores from applicants after dropping the requirement during the COVID-19 pandemic. Some scholars advocated for the testing requirement to be removed permanently – bringing up issues like systemic bias and racial inequality.
But the new research indicates that standardized tests are not biased and may even have the opposite effect.
Lead author Professor John Friedman, who teaches economics at Brown University, shared with The College Fix some of the details of his team’s work.
Regarding the research sample, Friedman said the team used data from first-year students at “multiple Ivy-plus colleges.”
“The choice to look at first year scores was entirely driven by data, in that we were able to collect comparable data on first year grades from our partner institutions but not for longer-term grades,” Friedman explained to The Fix in a recent interview. “We’ve looked at grades in later years of college within individual schools, and the broad pattern (test scores are predictive, high school GPA is not) remains.”
“In a separate paper, I have also shown that test scores but not high school GPA predict students’ post-college outcomes, including earning higher incomes, attending elite graduate schools, and working at prestigious firms,” the professor said.
“So while this is a limitation for this specific paper, I’m not concerned that the effects would be very different if we looked at longer-term outcomes,” he said.
As for the study’s findings on “race, gender, and socioeconomic status,” Friedman told The Fix:
“We are measuring the predictive power of both test scores and GPA among students with the same race, gender, and socioeconomic status … Importantly, this is not about comparing test scores and grades between students from different racial group groups, genders, or classes.”
Adam Kissel, a visiting fellow in education policy at the Heritage Foundation and former U.S. Department of Education official, also spoke with The Fix about the implications of Friedman’s team’s findings.
“Standardized tests designed to test college aptitude actually do predict college success—without bias,” Kissel wrote recently on X, linking to the working study.
In a recent interview with The Fix, Kissel elaborated, “This study shows that standardized tests are highly predictive of college success at more selective colleges. Selective colleges that went test-optional have been learning that they have done a worse job at admitting students who are likely to succeed, and this study helps show why.”
Concerning colleges, Kissel advised, “Highly selective colleges that want their students to succeed should heavily rely on standardized test scores as a first approximation of scholastic aptitude, regardless of group identity.”
With respect to prospective students, Kissel’s advice was specific to the selectivity of the school.
“This study notes other literature suggesting that test scores are less predictive of success at less-selective colleges,” he said. “Going test-optional is unwise for selective universities but not as big a mistake for other ones.
“Students should share their test scores, especially when applying to selective colleges, unless the scores are very low compared with published test score ranges or with the student’s GPA,” Kissel said.
It will be a little while yet before the full results of the research will be published.
As Friedman told The Fix, “Our paper (or, to be precise, a slightly shorter version of it given space constraints in the publication) will be published in the AEA Papers and Proceedings for the 2025 Annual Conference, which (I think) will come out in May.”
