Title:  Modeling non-linear growth using SEM: Structured Latent Curve Models.

 

Abstract: Structured latent curve models (Browne & du Toit, 1991; Browne, 1993) allow us to extend the classic SEM growth model to situations in which growth in a given construct is non-linear. Many of the developmental growth processes studied by psychologists are inherently non-linear. Language acquisition, learning to read, and memory decline as we age are all inherently non-linear growth processes. Psychologists also routinely employ measures which assess growth within a single testing session. For example, tests of memory, such as the California Verbal Learning Test and the Rey Auditory Verbal Learning Test, specifically assess an individual’s capacity to encode new information by testing how much new information is retained over a series of trials administered within the same testing session. The increase in the amount of new information retained over successive trials is also non-linear. 

I first explain how structured latent curve models are specified within the SEM framework. I then show how to implement these models using Mplus. The example presented uses data from the Alzheimer’s Disease Neuroimaging Initiative. Specifically, I will demonstrate how to use a structured latent curve model to model the increase in the number of words retained over the five learning trials of the Rey Auditory Visual Learning Test.

Presentation Slides and Materials