Performance Differences With and Without Differential Item Functioning in the Post Graduate Admission Test in Saudi Arabia Based on Gender and Ability Level

Eqbal Z, Darandari^*Muna A. Almeri

• King Saud University, Riyadh, Saudi Arabia

This study aimed to investigate Differential Item Functioning (DIF) based on gender and ability level for Post-Graduate General Aptitude Test (PGAT) items in Saudi Arabia, using classical methods (MH χ2, MH-LOR, BD χ2, and CDR). The study samples consisted of (4000) students distributed equally between males and females. For overall sample, 56 (54%) out of 104 items showed DIF: with (48%) of them favoring females and (41%) favoring males. For high ability sample, percentage of DIF items decreased across subtests, particularly for verbal sub-test. DIF items favoring females decreased (to 40%) and the ones favoring males increased (to 55%). ANOVA results showed that for overall sample, females outperformed males on total score and verbal ability, while males outperformed females on quantitative and logical abilities, significantly (p<.01). When DIF items were removed for overall sample, gender gap was reduced except for verbal ability, favoring females. For high ability sample, differences on total and sub-scores were not statistically significant except for quantitative ability, that favored males (p<.01). When DIF items were removed for high ability sample, gender differences were not statistically significant (p>.05). Thus, it was recommended to conduct stratified DIF analysis for ability admission tests based on ability area and level, gender and their interaction; and to report DIF size and direction for ability groups based on cut scores.