Investigating the Evidence of Differential Item Functioning From Interactive Process Log Data (Poster 21)

• In Event: AERA Poster Session Three
  In Poster Session: Measurement Theory and Application 2

Abstract

The advanced development of technology in digital-based assessment environments may introduce irrelevant variance in item responses and threats to item fairness in measurement. Hence, it is essential to assess differential item functioning (DIF) difference between subgroups. However, the item response data often contains missing data that can cause biased outcomes in DIF detection. To accurately detect DIF items for the new items, we investigated the association between informative process log features and DIF items, providing a framework for applying supervised machine learning methods to a set of features to precisely detect DIF items.

Authors

• Ziying Li, University of Florida
  Presenting Author

• Jinnie Shin, University of Florida
  Presenting Author

• Corinne Huggins-Manley, University of Florida
  Non-Presenting Author