Dahlke · The Journal of applied psychology 2019 · Retrospective cohort study · n=363,004

Effects of range restriction and criterion contamination on differential validity of the SAT by race/ethnicity and sex.

Cited 15 times in the scientific literature.

Level 3 - non-randomized controlled study

Level 3 by design analogy (large multicenter retrospective cohort study).

PubMed 30640487 · doi:10.1037/apl0000382 · record verified 2026-08-26

What was done

Researchers evaluated how multivariate range-restriction corrections and adjustments for criterion contamination (individual differences in course-taking patterns) affect the predictive validity of SAT scores across racial/ethnic groups and sex. The analysis utilized academic records from 363,004 students across 107 U.S. higher education institutions, evaluating predictive validity for both first-year and 4-year cumulative college GPAs.

What was found

The abstract reports statistical comparisons and directions without specific numerical correlation values. Controlling for course-taking patterns increased SAT validity coefficients compared to unadjusted correlations with obtained GPAs. After correcting for course-taking patterns and differential range restriction, SAT predictive validity for first-year GPA did not significantly differ between Black or Hispanic students and White students, but a significant Black-White difference remained for 4-year cumulative GPA. Validity estimates for both first-year and 4-year cumulative GPA were significantly lower among Asian students than White students after corrections. Observed validity was substantially lower for males than females, and adjusting for course-taking patterns increased the male-female validity difference.

Why it matters

This study shows that standard GPA correlations can underestimate test validity due to range restriction and differential course-taking. Applying psychometric corrections eliminates first-year predictive validity differences between Black/Hispanic and White students, while other subgroup differences persist or widen.

Limits

The abstract reports no numeric validity coefficients, effect sizes, or confidence intervals. The sample is restricted to U.S. college matriculants who took the SAT, and unmeasured factors such as institutional selectivity, socioeconomic status, and non-academic variables were not addressed.

Cited by