A heteroscedastic hidden Markov mixture model for responses and categorized response times

Area of Science:

• Psychometrics
• Cognitive Psychology
• Statistical Modeling

Background:

• Existing mixture models for psychometric tests face challenges with response time distributions, variance assumptions, and independence of latent variables.
• These limitations can bias results, especially when issues like heteroscedasticity or related processes occur simultaneously.
• Addressing these challenges is crucial for accurately understanding within-subjects differences in cognitive processes.

Purpose of the Study:

• To propose a novel statistical model that overcomes the limitations of existing mixture modeling approaches for psychometric tests.
• To develop a unified model capable of handling parametric distribution violations, heteroscedasticity, and related latent variables in response times.
• To demonstrate the practical utility and performance of the proposed model in analyzing cognitive processes.

Main Methods:

• Development of a heteroscedastic hidden Markov mixture model for analyzing both responses and categorized response times.
• The model integrates solutions for parametric distribution assumptions, unequal variances (heteroscedasticity), and dependent latent variables.
• Validation through a simulation study assessing parameter recovery and model resolution, followed by application to real-world data (WAIS-IV block design).

Main Results:

• The simulation study confirmed acceptable parameter recovery and the model's ability to distinguish between various special cases.
• The proposed heteroscedastic hidden Markov mixture model effectively addresses simultaneous challenges often present in response time data.
• Application to the WAIS-IV block design subtest demonstrated the model's practical applicability in psychological assessment.

Conclusions:

• The proposed heteroscedastic hidden Markov mixture model offers a robust and comprehensive approach to analyzing psychological processes in psychometric testing.
• This model provides a significant advancement by simultaneously addressing critical limitations of previous mixture modeling techniques.
• The findings support the model's utility for accurate and nuanced insights into individual cognitive differences.