Investigating Gender Differential Item Functioning Across Countries and Test Languages for PISA Science Items

摘要

This study uses PISA cycle 3 field trial data to investigate the relationships between gender differential item functioning (DIF) across countries and test languages for science items and their formats and the four other dimensions defined in PISA framework: focus, context, competency, and scientific knowledge. The data used were collected from 60 test language groups by 50 participating countries with a total of about 83,000 fifteen-year-old students. An IRT method is used to detect the gender uniform DIF for each of the language groups and the whole international sample. Gender directions have been drawn for each of the five item characteristic dimensions, for example, multiple choice and closed response items tend to favor males. The study also shows the effects of the country and test language on gender DIF in the international data. Findings from this study provide a potentially valuable contribution to the development of tests for international use.