An Exploratory Strategy to Identify and Define Sources of Differential Item Functioning

Area of Science:

• Psychometrics
• Educational Measurement
• Statistical Modeling

Background:

• Differential item functioning (DIF) sources are typically identified via qualitative expert review.
• Expert reviews may miss DIF sources outside their experience, leading to misidentification.
• Existing quantitative methods require prior knowledge of all potential DIF sources.

Purpose of the Study:

• To propose an exploratory strategy combined with the multiple-indicator multiple-cause (MIMIC) method to identify and name novel DIF sources.
• To address the limitations of current quantitative DIF analysis methods that assume all sources are known beforehand.

Main Methods:

• Developed an exploratory strategy integrating the MIMIC method for DIF source identification.
• Investigated the strategy's performance using simulation studies.
• Applied the strategy to real-world data for practical demonstration.

Main Results:

• The exploratory MIMIC method accurately identified the number and nature of DIF sources when DIF-free items defined the main dimension.
• The method successfully recovered items associated with each identified DIF source.
• Real data analysis confirmed the practical applicability of the proposed strategy.

Conclusions:

• The proposed exploratory MIMIC strategy offers an effective approach for discovering unknown DIF sources in item response data.
• This method enhances the efficiency and precision of the item review and revision process.
• It provides a valuable tool for psychometricians and test developers seeking to improve assessment quality.