Building Latent Class Regressions with Covariate Effects on Underlying and Measured Variables

Abstract

In recent years, latent class models have proven useful for analyzing relationships between measured multiple indicators and covariates of interest. Such models summarize shared features of the multiple indicators as an underlying categorical variable, and the indicators' substantive associations with predictors are built directly and indirectly in unique model parameters. In this talk, I aim to provide a detailed study on the theory and methods of building models that allow mediated relationships between primary predictors and latent class membership, but also allow direct effects of secondary covariates on the indicators themselves. Theory for model identification is developed. I propose an EM algorithm for parameter estimation and provide standard error estimates for parameters. Practical methods are developed to diagnose the adequacy of model. An analysis of how measured health impairments affect older persons' functioning is used for illustration. This model may improve researchers' tools for disentangling mechanistic influences on perceived health from risk factors.