Co-occurring dominance and ideal point processes: A general IRTree framework for multidimensional item responding

Area of Science:

• Psychometrics
• Cognitive Psychology
• Educational Measurement

Background:

• Rating scale item responses are influenced by both the trait measured and personal characteristics.
• Item Response Theory (IRT) Tree (IRTree) models analyze responses as sequential sub-decisions.
• Existing IRTree models primarily focus on dominance processes, limiting the analysis of other response rationales.

Purpose of the Study:

• To introduce a novel multidimensional IRT model (DI-MIRT) for IRTree models.
• To integrate co-occurring dominance and ideal point processes within IRTree sub-decisions.
• To expand the flexibility and application scope of IRTree models for analyzing cognitive item responding.

Main Methods:

• Developed the DI-MIRT parameterization for IRTree models.
• Incorporated multiple dominance and ideal point processes, and their combinations, into sub-decisions.
• Conducted a simulation study to evaluate parameter recovery and model accuracy.

Main Results:

• The DI-MIRT parameterization accurately reflects co-occurring dominance and ideal point processes.
• IRTree models with DI-MIRT parameterization demonstrated excellent parameter recovery.
• A significant advantage was observed compared to traditional IRTree models with purely sequential processes.

Conclusions:

• The DI-MIRT model offers a flexible framework for parameterizing IRTree sub-decisions.
• This new model expands the theoretical assumptions applicable to cognitive item responding.
• The DI-MIRT model enhances the analysis of response styles and improves measurement in real-world applications.