Test equating in the presence of DIF items

Area of Science:

• Psychometrics
• Educational Measurement
• Statistical Modeling

Background:

• Differential item functioning (DIF) can compromise the accuracy of test equating.
• Existing methods for controlling DIF in equating may not be fully effective.
• Accurate and stable parameter estimates are crucial for valid test equating.

Purpose of the Study:

• To propose and evaluate a novel multilevel measurement model for test equating that explicitly controls for DIF effects.
• To assess the accuracy and stability of item and ability parameter estimates generated by the proposed model.
• To compare the performance of the proposed model against traditional multiple-group concurrent equating designs.

Main Methods:

• Development of a multilevel measurement model incorporating DIF control.
• Utilizing randomly simulated data to examine model performance.
• Comparison of the proposed model with two multiple-group concurrent equating designs: one replacing DIF items and one retaining DIF items without control.

Main Results:

• The proposed multilevel measurement model generally outperformed the two comparison models in terms of accuracy and stability of parameter estimates.
• The model's effectiveness was observed across most simulated conditions.
• The proposed model provides a more robust approach to handling DIF in test equating.

Conclusions:

• The multilevel measurement model offers a significant advancement in controlling for DIF in test equating.
• This approach enhances the reliability and validity of equating procedures.
• The proposed model is recommended for applications requiring precise measurement and fair comparisons across different test forms.