Differential item functioning detection across multiple groups

Area of Science:

• Psychometrics
• Educational Measurement
• Item Response Theory (IRT)

Background:

• Differential item functioning (DIF) analysis requires converting item parameters to a common metric across groups.
• Simultaneous scale conversion and DIF detection are challenging due to their interdependent nature.

Purpose of the Study:

• To propose a novel method for simultaneously performing scale conversion and DIF detection.
• To develop an approach where scale conversion automatically accounts for DIF.

Main Methods:

• A novel method integrating scale conversion and DIF detection.
• Penalized likelihood estimation for automatic selection of DIF items.
• Simultaneous estimation of item parameters and scale conversion factors.

Main Results:

• The proposed method effectively integrates scale conversion and DIF detection.
• Automatic selection of DIF items using penalized likelihood estimation.
• Real-data and simulation studies demonstrate the method's good performance.

Conclusions:

• The novel simultaneous approach improves the accuracy of DIF detection and scale conversion.
• This method provides a more robust way to analyze group differences in item parameters.
• The findings have implications for fair and accurate assessment across diverse populations.