B0305
Title: Effect size measures for differential item functioning
Authors: Daphna Harel - New York University (United States) [presenting]
Abstract: Differential item functioning (DIF) occurs when groups (such as demographic groups like gender or age) endorse a given item on a multi-item scale with different probabilities, even after the overall scale score is controlled for. However, traditional metrics for detecting DIF are based on statistical significance, which is known to be a function of sample size. When large datasets are used in DIF analyses, many items are expected to be flagged as having DIF, even when the magnitude or impact of this DIF is small. Different effect size measures are explored for DIF to assess the impact of DIF on factor scores arising from a multi-item questionnaire.
