Integrating person-centered and variable-centered analyses: growth mixture modeling with latent trajectory classes

Area of Science:

• Quantitative Psychology
• Substance Abuse Research
• Developmental Trajectories

Background:

• Alcohol research often requires person-centered approaches to identify heterogeneous groups, such as those susceptible to alcohol dependence.
• Understanding developmental trajectories in longitudinal data benefits from a person-centered perspective.
• Recognizing heterogeneity in alcohol, drug, and mental health research has spurred theories of multiple developmental pathways.

Purpose of the Study:

• To provide an overview of novel methods integrating variable- and person-centered analyses.
• To present a general latent variable modeling framework accommodating both continuous and categorical latent variables.
• To illustrate the application of these integrated methods in alcohol research.

Main Methods:

• Latent class analysis (LCA)
• Latent transition analysis (LTA)
• Latent class growth analysis (LCGA)
• Growth mixture modeling (GMM)
• General growth mixture modeling (GGMM)
• These methods are framed within a general latent variable modeling framework.

Main Results:

• Illustrative examples using the National Longitudinal Survey of Youth (NLSY) data.
• Latent class analysis identified four distinct classes of antisocial behavior.
• Four distinct heavy drinking trajectory classes were identified.
• The study examined relationships between latent classes, background variables, and consequences.

Conclusions:

• Novel methods successfully integrate previously distinct variable- and person-centered analytical approaches.
• The general latent variable framework facilitates the combination of diverse models.
• This integrated approach stimulates new research questions addressing both person- and variable-centered aspects of complex phenomena.