Differential item functioning /
Steven J. Osterlind, Howard T. Everson.
- 2nd ed.
- x, 87 s. : ill. ; 22 cm.
- Quantitative applications in the social sciences ; 161 .
Revised ed. of: Test item bias.
Introduction: Audience and requisite background. Role of DIF investigation. Fairness and DIF. Bias and DIF. DIF and discrimination. Understanding the term DIF. Working vocabulary for this monograph. -- Description of DIF: Formal definition of DIF. DIF: Uniform or not -- Statistical facets of DIF: Not mean (average) difference. Error in measurement. Differences are systematic. Matching on ability or proficiency. Conditioning criterion: internal versus external. Purification of data -- Important considerations: DIF: statistics versus reasoned judgment. statistical bias is not unfairness. Whole test versus individual items. Number: items and sample. A sensible perspective -- History of test bias and DIF: Standards for test fairness -- Quick-but-incomplete methods: Ordinal ranking of items. Ability group methods. Outdated ANOVA methods -- Mantel-Haenszel procedure: CHi-Square contingency table. M-H odds ratio -- Nonparametric methods: DIF with SIBTEST. Dorans’s standardization -- IRT-based methods: IRT framework. Item response curves. The one-parameter model. The two-parameter model. The three-parameter model. DIF methods based on IRT. Differences in item parameters. Likelihood ratio test. Area measures. IRT approaches for detecting DIF in polytomous items -- Logistic regression: The logistic regression DIF expression. Example of DIF by logistic regression -- Specialized scored items. Computer-based testing. Computer-adaptive testing. Translated tests -- Future Directions: Validity argument. Null hypothesis testing. Statistical modeling (HLM and Other). Equivalence testing. DFIT and CDIF testing.