Identifiability of Latent Class Models with Covariates

Area of Science:

• Statistics
• Psychometrics
• Educational Measurement

Background:

• Latent class models with covariates are foundational in psychological, social, and educational research.
• The fundamental issue of parameter identifiability in these models remains a significant challenge.
• Prior work by Huang and Bandeen-Roche (2004) explored local identifiability but did not provide sufficient conditions.

Purpose of the Study:

• To address the unresolved identifiability issue in latent class models with covariates.
• To establish conditions ensuring global identifiability of model parameters.
• To extend identifiability results to polytomous-response cognitive diagnosis models (CDMs) with covariates.

Main Methods:

• Analysis of local identifiability conditions for latent class models with covariates.
• Development of new theoretical conditions for global identifiability.
• Generalization of findings to cognitive diagnosis models (CDMs) with covariates.

Main Results:

• Demonstrated that existing local identifiability conditions are necessary but not sufficient.
• Established conditions for ensuring the global identifiability of latent class model parameters.
• Extended these identifiability results to polytomous-response CDMs with covariates.

Conclusions:

• The established conditions provide a robust framework for ensuring parameter identifiability in latent class models with covariates.
• This work advances the theoretical understanding of latent class models and their applications, particularly in cognitive diagnosis.
• The findings offer crucial guidance for researchers using these models in complex data analysis scenarios.