A Comparison of Two Differential Item Functioning Methods for Analyzing Rasch Model Data: A Monte Carlo Investigation

• In Event: AERA Poster Session 10
  In Poster Session: Applications of the Rasch Measurement Model

Abstract

One of the key components in educational tests is to detect biased or Differential Item Functioning (DIF) items for the evaluation of fairness and validity. A Monte Carlo simulation study was conducted to compare the performance of the newly-developed item-focused trees procedure and the popular between-fit statistic in the Rasch dichotomous model. A total of three conditions were varied (i.e. number of examinees per group, percent/number of items exhibiting DIF, and the magnitude of DIF presence). Results showed that item-focused trees generally outperformed the between-fit statistic and optimally when both sample size and magnitude of DIF were large. However, the between-fit statistic performed well when sample size was small and only 2 out of 20 items exhibited DIF.

Authors

• Ning Jiang, University of South Carolina - Columbia

• Kelvin Terrell Pompey, University of South Carolina

• Yin Burgess, National Registry of EMT's

• Tiejun Zhang, University of South Carolina