A new method for detecting differential item functioning in the Rasch model

摘要: Differential item functioning (DIF) can lead to an unfair advantage or disadvantage for certain subgroups in educational and psychological testing. Therefore, a variety of statistical methods has been suggested detecting DIF the Rasch model. Most these are designed comparison pre-specified focal reference groups, such as males females. Latent class approaches, on other hand, allow detect previously unknown groups exhibiting DIF. However, this approach provides no straightforward interpretation with respect person characteristics. Here we propose new method detection based model-based recursive partitioning that be considered compromise between those two extremes. With it is possible subjects DIF, which not prespecified, but result from combinations observed covariates. These directly interpretable thus help understand sources DIF. The background construction first introduced by means instructive example, then applied data general knowledge quiz teaching evaluation.