A Mixed Sequential IRT Model for Mixed-Format Items.

Area of Science:

• Educational Measurement
• Psychometrics
• Cognitive Psychology

Background:

• Traditional item response models struggle with mixed-format items (e.g., multiple-choice and open-ended).
• Sequential response processes and scoring require specialized modeling approaches.
• Existing polytomous models like GRM, GPCM, and SRM have limitations in capturing complex item structures.

Purpose of the Study:

• To propose three novel mixed sequential item response models (MS-IRMs) for mixed-format items.
• To enhance the analysis of sequential response processes and cognitive processes in assessments.
• To improve upon conventional polytomous models by incorporating task-specific processing functions.

Main Methods:

• Development of three distinct mixed sequential item response models (MS-IRMs).
• Utilized simulation studies to evaluate model performance, including parameter recovery and model fit.
• Applied the proposed MS-IRMs to real data from TIMSS 2007 for comparative illustration.

Main Results:

• All three proposed MS-IRMs demonstrated superior performance over traditional models (SRM, GRM, GPCM).
• The MS-IRMs showed improved parameter recovery.
• Enhanced model fit was observed for the proposed MS-IRMs compared to existing methods.

Conclusions:

• The developed MS-IRMs offer a more effective approach for analyzing mixed-format items with sequential response processes.
• These models provide better insights into individual response and cognitive processes.
• MS-IRMs represent a significant advancement over conventional polytomous and sequential models in educational measurement.