A sequential exploratory diagnostic model using a Pólya-gamma data augmentation strategy

Area of Science:

• Educational Measurement
• Psychometrics
• Cognitive Psychology

Background:

• Cognitive diagnostic models (CDMs) classify individuals into latent proficiency classes (attribute profiles).
• Previous work adapted Pólya-gamma data augmentation for binary response CDMs using logistic item response functions and Bayesian Gibbs sampling.
• Existing methods often focus on binary outcomes, limiting applications for ordinal data.

Purpose of the Study:

• To propose a novel sequential exploratory diagnostic model tailored for ordinal response data.
• To extend the Pólya-gamma data augmentation strategy to handle ordinal response processes within CDMs.
• To develop efficient estimation methods for this extended model.

Main Methods:

• Developed a sequential exploratory diagnostic model with logit-link parameterization at the category level for ordinal data.
• Extended the Pólya-gamma data augmentation strategy to accommodate ordinal response processes.
• Implemented a Gibbs sampling procedure for efficient Markov chain Monte Carlo (MCMC) estimation.

Main Results:

• The proposed model effectively handles ordinal response data in cognitive diagnostic assessments.
• The extended Pólya-gamma augmentation strategy proved efficient for ordinal data.
• Monte Carlo simulations demonstrated the model's performance and stability.

Conclusions:

• The sequential exploratory diagnostic model offers a valuable advancement for analyzing ordinal data in educational and psychological assessments.
• The extended Pólya-gamma augmentation strategy provides an efficient computational approach for complex CDMs.
• This research facilitates more nuanced understanding of individual proficiencies through attribute profiling.