Abstract
Influence of iterative linking on differential item functioning detection using Raju's area measures and Lord's statistic. In order to improve the correct identification of test items which show differential functioning (DIF), several test purification procedures have been proposed. The present study uses a bietapic purification procedure (Hidalgo & López, 1997b) to improve DIF detection with Lord's χ2 statistic and Raju's area measures. Sample size (250, 500 and 1.000 subjects), amount of DIF (0.4 and 0.6), types of DIF —uniform, non-uniform and mixed— and percent of DIF test items (10% and 30%) were manipulated. The results show that the bietapic procedure reduces the false positive rates.
