Gender Differential Item Functioning (GDIF) in an Online Intelligence Test

摘要

This study explores the differential performances between genders among Malaysian public university students. The instrument used was Malaysian intelligence test known as UKM2. The study aims at measuring gender bias in the UKM2 on-line test by investigating differential item functioning (DIF) across genders. There were 124 male and 373 female students who were randomly selected from one of the public universities in Malaysia as samples in the study. The Malaysian intelligence test consists of four indices (a) Verbal Comprehension Index (VCI), (b) Perceptual Reasoning Index (PRI), (c) Working Memory Index (WMI) and (d) Processing Speed Index (PSI). The overall 370 items were made up of dichotomous and partial credits from 13 constructs. Data gathered were analyzed using Winsteps version 3.64.2, a Rasch-based item analysis program. Study shows that gender-related DIF (GDIF) exists in the test administered. GDIF contrast which favored the males and females are displayed in 77 items and 76 items respectively. However, only 29 items are exhibiting significant GDIF, based on the size ranging from 0.61 to 5.61 (logits). Males excelled in two indices namely VCI and WMI, while females in PRI and PSI indices. The authors discuss possible explanation for differential clustering tendencies of items displaying GDIF and some practical and theoretical implications of GDIF in the development and interpretation of the online test.