Reasons for Gender-Related Differential Item Functioning in a College Admissions Test

• In Event: 25.046 - Large-Scale Assessments in International Settings: Methods and Practices

Abstract

The SweSAT is a Swedish large scale college admissions test and in this study differential item functioning (DIF), focusing on gender differences, was examined in 400 verbal and 400 quantitative items from five test administrations using real data from 250,000 test takers. The Mantel-Haenszel and logistic regression procedures were used to analyze the data for uniform and non-uniform DIF. They had a high matching percentage of their results and found mainly negligible DIF in the quantitative items and some moderate/intermediate and large DIF in the verbal items with about half of the DIF favoring each gender. These results provide empirical evidence that there is very little gender bias in the SweSAT despite large observed score differences.

Author

• Jonathan Wedman, Umeå Universitet