Nidhi Kohli, University of Minnesota
Bayesian Longitudinal Mediation Models
We present two interrelated studies introducing Bayesian (non)linear random effects mediation models (B(N)REMM). In the first study, we develop both intrinsically linear and nonlinear forms of B(N)REMM to directly estimate longitudinal mediation effects, addressing key limitations of existing structural equation modeling (SEM) approaches. Specifically, we propose: (1) a linear trend model (L-BREMM), and (2) a piecewise linear-linear model with random changepoints (P-BREMM) to capture segmented trends. Using data from the Early Childhood Longitudinal Study–Kindergarten Cohort (ECLS-K), we also investigate the consequences of omitting confounders in nonlinear mediation models.
To account for heterogeneity in mediation processes, we extend the framework to Bayesian (nonlinear) growth mixture mediation models (B(N)GMMM), which allow for subgroup-specific mediation pathways and treatment effects of an intervention variable X on a longitudinal outcome Y, mediated by M. We evaluate model performance and robustness through a comprehensive Monte Carlo simulation study and illustrate the methods using a real-world clinical trial intervention dataset.
about the speaker
Dr. Nidhi Kohli is the Royal and Virginia Anderson Professor of Quantitative Methods in Education in the Department of Educational Psychology at the University of Minnesota. She is also the Director of the Longitudinal Methods Development (LMD) Lab, where she and her students develop innovative statistical methodologies to analyze complex nonlinear longitudinal data structures and assess model fit for these approaches. Dr. Kohli’s research has been supported by the National Institutes of Health (NIH), the National Science Foundation (NSF), and the Institute of Education Sciences (IES). She serves as the Editor of the Application and Case Studies section of Psychometrika and is a member of the editorial boards for the Journal of Educational and Behavioral Statistics, Multivariate Behavioral Research, Psychological Methods, and Educational and Psychological Measurement.