Recent advances in analysis of differential item functioning in health research using the Rasch model

Area of Science:

• Psychometrics
• Health outcomes research
• Statistical analysis

Background:

• Rasch analysis and Differential Item Functioning (DIF) are vital for assessing health outcome measures.
• Splitting DIF items is a common technique to maintain measurement precision.
• Distinguishing real from artificial DIF is crucial for accurate analysis.

Purpose of the Study:

• To present recent advances in DIF analysis within a unified methodology.
• To focus on using Analysis of Variance (ANOVA) for detecting uniform and non-uniform DIF.
• To illustrate DIF identification and handling in health research using the Rasch model.

Main Methods:

• Rasch analyses of DIF were performed on psychosomatic problem measures.
• Swedish data from the Health Behaviour in School-aged Children study (grades 9, 1985-2014) were used.
• Analysis of variance (ANOVA) of residuals was employed to detect DIF.

Main Results:

• ANOVA efficiently identified DIF, with magnitude quantified.
• Resolving items by group and using test equating accounted for DIF.
• Real DIF was shown to impact person measurement estimates.

Conclusions:

• Distinguish real from artificial DIF by resolving items sequentially, not simultaneously.
• Resolving or deleting DIF items can impact content validity; consider the source of DIF.
• Decisions on handling DIF should incorporate external information to preserve instrument validity.