Description

Latent class modeling refers to a group of techniques for identifying unobservable, or latent, subgroups within a population. Researchers have developed and expanded methods like latent class analysis (LCA) and latent transition analysis (LTA) over the last two decades. Our current research focuses on expanding methods to include latent class variables in larger models of complex developmental processes. Latent class analysis (LCA) identifies unobservable subgroups within a population. We work to expand LCA models to allow scientists to better understand the impact of exposure to patterns of multiple risks, as well as the antecedents and consequences of complex behaviors, so that interventions can be tailored to target the subgroups that will benefit most. Latent transition analysis (LTA) is a related method that allows scientists to estimate movement between subgroups over time.

LCA example

LCA Introductory Example: Profiles of Teen Sex and Drug Use

In this example, LCA identifies five subgroups of teenagers based on their substance use and sexual behaviors. The latent variable “youth risk behavior” is measured by the observed variables “sex,” “drinking,” “smoking,” and “other drugs.” This analysis allows us to identify complex behavior patterns and variables that predict high-risk behavior patterns, as well as identify subgroups of youth who are at-risk for negative health consequences. With this information, scientists can develop interventions that target individuals with the greatest need.

Read more here.

Resources

There are a variety of resources available to help you learn more about LCA. See our Resources page for the following:

Static Models

LCA: Baseline LCA with 3+ level categorical indicators

This code fits a longitudinal latent class model, using categorical indicators with 3+ levels, to identify latent classes indicated by multidimensional experiences of racism and heterosexism during the transition to adulthood among sexual minority men of color.

View Model

LCA: Latent class moderation

This code demonstrates how to use a latent class moderator to examine heterogeneity in intervention effects among adolescents receiving treatment for cannabis use. First, the code identifies latent classes of contextual and individual risk at baseline using LCA. Then, it uses an adjusted 3-step approach with BCH weights to regress the outcomes on level of care, latent class membership, the interaction between them, and covariates.

View Model

LCA: LCA with a grouping variable and without measurement variance

This code fits a 4-class, latent-class model for marijuana use and attitudes using 7 binary indicators of the latent class variable. It includes a grouping variable for year, and observations came from 3 different years. Measurement invariance across groups is not imposed resulting in an unrestricted latent class model with multiple groups.

View Model

Let’s stay in touch.

We are in this together. Receive an email whenever a new model or resource is added to the Knowledge Base.