Causal Inference in Latent Class Analysis.

Area of Science:

• Social Sciences
• Behavioral Sciences
• Health Sciences

Background:

• Latent class analysis (LCA) is valuable for identifying unobserved subgroups.
• Causal inference methods are needed to determine factors influencing membership in these subgroups.

Purpose of the Study:

• To integrate causal inference techniques with latent class analysis (LCA).
• To demonstrate propensity score methods (matching and inverse propensity weighting) for causal inference in LCA.
• To estimate the causal effect of college enrollment on adult substance use latent class membership.

Main Methods:

• Propensity score matching and inverse propensity weighting were applied to LCA.
• Multiple imputation was used for handling missing data on confounders, exposure, and outcome.
• Analysis utilized data from the National Longitudinal Survey of Youth 1979.

Main Results:

• The study successfully estimated the causal effect of college enrollment on adult substance use latent classes.
• Demonstrated the practical application of propensity score methods within an LCA framework.

Conclusions:

• Integrating causal inference with LCA provides powerful tools for social, behavioral, and health research.
• Propensity score techniques enable robust estimation of treatment effects on latent class membership.