Differential Validity and Differential Prediction for a Graduate School Admission Test

• In Event: Poster Session 19
  In Poster Session: Test Validity Research and Evaluation SIG: Poster Presentation

Abstract

Because group differences are often observed on standardized tests, the fairness of using scores for admissions selection is often questioned. Validity studies are one way to provide evidence for the fair use of scores. The current study examines differential validity and differential prediction for a graduate school admission test used across the world. Based on nearly 6,000 cases, results suggested that validity evidence for predicting performance in graduate school is relatively consistent across groups. Patterns of overprediction for female students, under-represented minorities, non-native English speakers, and certain citizenship groups suggested that these groups were not disadvantaged by the use of the admission test scores. Further research should include larger sample sizes for more robust comparisons and potentially moderating variables.

Author

• Eileen Talento-Miller, The Graduate Management Admission Council