glca: An R Package for Multiple-Group Latent Class Analysis

Area of Science:

• Statistics
• Psychometrics
• Social Sciences

Background:

• Latent Class Analysis (LCA) is used to identify unobserved subgroups within a population.
• Understanding group similarities and differences is crucial in LCA, but complex data structures can pose challenges.
• Measurement invariance tests are key to distinguishing identical versus differing latent structures across groups.

Purpose of the Study:

• To develop an R package, glca, for exploring and testing differences in latent class structures across populations.
• To accommodate multilevel data structures within latent class analysis.
• To provide statistical procedures for comparing latent class models between groups.

Main Methods:

• Implementation of fixed-effect LCA for populations segmented by observed group variables.
• Implementation of nonparametric random-effect LCA for situations with numerous group levels, identifying group-level latent variables.
• Development of statistical tests within the glca package for exploring group differences in various LCA models.

Main Results:

• The glca package offers a comprehensive framework for multilevel latent class analysis.
• It provides distinct approaches for fixed-effect and random-effect LCA, enhancing flexibility.
• The package facilitates robust statistical testing of measurement invariance and latent structure differences across groups.

Conclusions:

• The glca R package provides valuable tools for researchers investigating population heterogeneity using LCA.
• It effectively addresses the complexities of multilevel data in latent class modeling.
• The package supports rigorous examination of group differences in latent structures, advancing comparative research.