Biostatistics and Bioinformatics Core Training Seminar: "Flexible Structural Equation Models with Longitudinal Data"

Note: Due to a sound board problem, the introductory sound was lost. The sound is fixed at 3 minutes.

Chris Slaughter, Ph.D., Associate Professor of Biostatistics, presented a lecture titled "Flexibly Structural Equation Models with Longitudinal Data" on Monday, Apr. 10, 2017, from 12:00-1:00 p.m. in Room 241 of the Vanderbilt Kennedy Center.

Structural equation modeling can provide a framework for testing associations among observed variables and latent constructs in longitudinal studies. However, the models rely on underlying assumptions that can affect the validity of conclusions. In this talk, Dr. Slaughter will discuss assumptions about the error distribution and directionality of effect in a generalized growth mixture model. The model utilizes repeated measurements on a subject to estimate growth trajectories, predictors of class membership, and associations with future outcomes. He discussed a more flexible approach to relax Normality assumptions by allowing latent factors to follow finite mixture distributions. Researchers apply their methods to a study of early fetal growth and pregnancy outcomes.

This lecture is co-sponsored by the Department of Biostatistics. For more information on the series, contact Julie Lounds Taylor.

Last Updated: 4/10/2017 2:08:29 PM

Go to the news and video index