Abstract
Data from a 45-item vocabulary test (tryout edition) were used to assess the degree of congruence between the signed area measure, the unsigned area measure, Lord's chi-square statistic, and the Mantel-Haenszel (MH) technique in identifying items with significant differential item functioning (DIF). In both the Black-White and Female-Male comparisons, the three methods based on item response theory (IRT) identified the same items as having statistically significant DIF. There was close agreement between the MH technique and the three IRT-based procedures only in the Female-Male comparison; different items were identified as biased in the Black-White comparison.
Original language | English (US) |
---|---|
Pages (from-to) | 301-314 |
Number of pages | 14 |
Journal | Educational and Psychological Measurement |
Volume | 53 |
Issue number | 2 |
DOIs | |
State | Published - Jun 1993 |
ASJC Scopus subject areas
- Education
- Developmental and Educational Psychology
- Applied Psychology
- Applied Mathematics