The standardized testing debate turns on two points: whether admissions tests predict school success, and whether they treat students fairly. The best large-scale evidence answers yes to prediction and splits on fairness. In Psychological Bulletin, Sackett and colleagues ran the key check. Controlling for family income moved the test-grade tie only from a .47 to a .44 correlation. A correlation of .44 still leaves most of each student's outcome unexplained.
That gap between solid prediction and partial explanation is where the honest middle sits. Tests do work as statistical signals. They do not seal a student's fate, and they carry real gaps across income and race groups. Admissions rules reach beyond tests as well, as the recent fight over the fixed-period F-1 visa rule showed for international students.
This explainer walks through the main published findings, their sample sizes, and their limits. Related coverage sits in the site's education-news hub.
Do Admissions Tests Predict School Success?
The strongest evidence is a 2009 meta-style analysis in Psychological Bulletin. Paul Sackett, Nathan Kuncel, and their team pooled multiple large data sets with test scores, grades, and family background. Test scores tracked later academic performance at a correlation of .47. Even after controlling for socioeconomic status, the tie fell only to .44. For related coverage, see Harvard's Fight Over Its SEVP Certification Reshaped Visa Politics.
The team's own reading was blunt. Critics say the tests merely measure family wealth. These data said otherwise: the vast majority of the test-grade relationship stood on its own. This is prediction, not proof of cause, and a .47 tie describes groups, not any single applicant.
How Large Is the Income Effect?
Family background does track scores. In the same paper, socioeconomic status related to SAT scores at a correlation of .42 among test takers. So the worry about wealth is not made up. The open question was whether that link explains away the test's predictive power, and in these data it did not.
A 2012 study in Psychological Science scaled the check up. Sackett, Kuncel, and colleagues examined 143,606 students across 110 colleges and universities. The SAT added predictive power beyond high school grades, and controlling for status did not change that in a major way. The bigger barrier for low-income students, it found, was entering admissions at all. Rejection by colleges was not the main block.
What About Claims of Test Bias?
The bias question has direct evidence from medicine. A review committee examined the MCAT from 2008 to 2011. The work, by Davis, Sackett, and colleagues, appeared in Academic Medicine. Black and Latino examinees' mean scores ran below white examinees' scores, a gap that mirrors other tests and applicant grades. On later medical school measures, though, the review found no evidence the MCAT was biased against black and Latino applicants.
The same report noted context that keeps the debate alive. Families of white, black, and Latino aspiring doctors differed in parental education and income. Admission committees accepted majority and minority applicants at similar rates. Gaps at entry, in that account, form before the test day arrives.
How Strong Is This Evidence?
- Prediction: pooled data sets, correlation of .47 with grades, replicated across 143,606 students.
- Fairness checks: MCAT review of four years of data, no prediction bias found on later performance.
- Open sides: causes of group gaps, and the weight tests should carry, remain argued.
The limits are structural. These are correlational studies, so they cannot prove that scores cause success. A .47 tie means test scores capture about a fifth of the variance in grades. The findings cover United States tests, chiefly the SAT and the MCAT. Grades also miss part of what colleges value.
Conclusion: Strong Signal, Unsettled Fight
The published record supports a narrow claim. Admissions tests predict later academic performance at useful strength. That prediction survives controls for family income. The record does not support calling the tests pure wealth proxies. It also does not settle fairness. Score gaps form early, and their roots reach past the exam room.
