This study examined the predictive validity of postgraduate entrance-test scores for cumulative grade point average (GPA) and study duration at the Postgraduate School of Universitas Terbuka, Indonesia. Using retrospective institutional records, the study included 112 alumni admitted under comparable testing procedures implemented from 2021 onward; analyses involving English proficiency were based on 111 cases because one score was recorded on a non-comparable scale. Pearson correlations and multiple linear regression were used to assess predictive relationships. Numerical reasoning showed the strongest bivariate association with GPA (r = .227, p = .016), followed by verbal reasoning (r = .195, p = .040). No entrance-test component was significantly associated with study duration. In the multivariable model, English proficiency (β = −.233, p = .011), numerical reasoning (β = .271, p = .017), and figural reasoning (β = −.286, p = .004) were significant predictors of GPA, with the model explaining 17.9% of its variance (adjusted R² = .139). However, after adjustment for study program, only numerical reasoning remained statistically significant (β = .230, p = .044). Entrance-test scores provided little evidence of predictive utility for study duration after program adjustment. Overall, the findings indicate limited and component-specific predictive validity, with program-level heterogeneity influencing the observed relationships. These results highlight the need to evaluate entrance-test components in relation to specific academic outcomes and institutional contexts rather than treating aggregate test performance as a uniformly reliable predictor of postgraduate success.
Copyrights © 2026