Effect of Purification Procedures on DIF Analysis in IRTPRO.

Area of Science:

• Psychometrics
• Educational Measurement
• Statistical Modeling

Background:

• Purification is a standard procedure to improve test performance by addressing differential item functioning (DIF).
• Traditional purification methods, involving reestimating ability parameters after removing DIF items, can be labor-intensive.
• IRTPRO 3 offers DIF analysis but lacks built-in purification functionalities.

Purpose of the Study:

• To investigate the effectiveness of two novel purification methods for differential item functioning (DIF).
• To evaluate the impact of these methods on statistical power and Type I error rates in DIF detection.
• To develop an accessible tool for implementing purification procedures.

Main Methods:

• A simulation study was designed to compare two new purification techniques.
• Item response theory (IRT) models were employed for the analysis.
• A web-based program was developed to facilitate the application of the purification procedure.

Main Results:

• One of the investigated purification methods demonstrated statistically significant improvements in both statistical power and Type I error control.
• The newly developed web-based program effectively streamlined the application of the purification procedure.
• The results indicate enhanced accuracy in identifying differential item functioning.

Conclusions:

• The novel purification procedure offers a more accurate and efficient approach to handling differential item functioning.
• The web-based tool democratizes access to advanced psychometric analysis techniques.
• This advancement aids researchers in developing more equitable and precise assessment instruments.