Join us for the Quantitative Psychology Colloquium with Dr. Johnny Zhang (University of Notre Dame)!
This event will take place on Zoom. Join using this link.
Title: Bayesian Methods for Handling Longitudinal Ceiling Data
Abstract: Longitudinal data are commonly collected in psychological and educational research to study change. If the instruments used have a fixed upper bound, it can lead to ceiling effects where observations are clustered at the maximum value. Analyses that ignore censoring can underestimate change and distort covariate effects. In this talk, I will present methods to handle ceiling effects in longitudinal studies for both continuous and discrete data. We show how to expand the Tobit regression model to account for censoring. Specifically, we link the observed ceiling data with an underlying latent process to explicitly model the censoring mechanism. Through both real data analysis and simulations, we demonstrate that our approach provides more accurate estimates of change and covariate effects compared to traditional methods that ignore censoring.