Minimum Distance Estimation of Multidimensional Diffusion-Based Item Response Theory Models.

Area of Science:

• Psychometrics
• Cognitive Psychology
• Statistical Modeling

Background:

• Diffusion-based item response theory (dIRT) models offer advanced measurement capabilities for psychological tests, analyzing both responses and response times.
• Current limitations exist in fitting multidimensional dIRT models due to computationally intensive estimation methods like marginal maximum likelihood and inefficient weighted least squares estimators.

Purpose of the Study:

• To propose a novel, more efficient, and less computationally demanding estimator for multidimensional diffusion item response theory models.
• To address the practical challenges in applying multidimensional dIRT models in psychological research.

Main Methods:

• Developed a minimum distance estimation approach.
• The proposed estimator models sample quantiles and sample covariances.
• Investigated estimator performance through a comprehensive simulation study.

Main Results:

• The simulation study demonstrated that the proposed minimum distance estimator performs well.
• The new estimator is more efficient than the weighted least squares method and less demanding than maximum likelihood estimation.
• The estimator's practical utility was confirmed through application to real-world data.

Conclusions:

• The proposed minimum distance estimator offers a viable and efficient alternative for fitting multidimensional diffusion item response theory models.
• This advancement facilitates broader application of dIRT models in psychological measurement and analysis.
• The method provides a computationally feasible approach for complex psychometric modeling.