Category: Addictive Behaviors
Keywords: Addictive Behaviors | Research Methods | Alcohol
Presentation Type: Symposium
Latent profiles of substance use data have been identified across a variety of domains. Typically, researchers have identified a number of variables to distinguish latent classes of substance users (e.g., heavy users, light users, abstainers) and then tested whether latent classes differ on a variety of outcome and/or predictor variables. While interesting, the field need not be limited to examining latent profiles based only on the values of variables. Structural Equation Modeling with Mixtures (SEMM) allows for identifying latent classes that differ based on patterns of relationships between variables (e.g., on all relations specified in a SEM). This allows researchers to examine characteristics that determine whether a particular construct is likely to be a mechanism of change or for whom direct effects between variables may be more or less strong (e.g., growth mixture modeling, latent transition analysis). An example of a SEMM will be presented in order to demonstrate the utility of this modeling strategy to differentiate relations among variables within latent classes. The demonstration will show how to construct latent classes of mediation paths. More specifically, latent classes of the mediation path from College-Related Alcohol Beliefs (CRAB) to Alcohol Use (AU), via Protective Behavioral Strategy (PBS) use will be shown. Analyses indicated that the 2-class solution had excellent classification quality (e.g., entropy = .97), and yielded clinically meaningful results. The first class, which composed the majority of the sample, had a significant indirect effect from CRAB-PBS-AU, whereas, in the second class, the indirect effect was not significant. This demonstration reveals that while PBS was a mechanism through which CRAB translates into AU for some individuals, it may be not for others. Identifying latent subgroups of individuals with differing relations among constructs has clinical utility. Analyses such as this one provide insights that clinicians can use to screen and treat clients more effectively, e.g., identifying characteristics of experiencing PBS as a mechanism of change or not.
Colorado State University
Friday, November 17
10:15 AM – 11:45 AM
The asset you are trying to access is locked. Please enter your access key to unlock.