Bayesian growth curve models with the generalized error distribution |
| |
Authors: | Zhiyong Zhang |
| |
Affiliation: | Department of Psychology , University of Notre Dame , 118B Haggar Hall, Notre Dame , IN , 46556 , USA |
| |
Abstract: | To deal with the longitudinal data with both leptokurtic and platykurtic errors, we extend growth curve models using the generalized error distribution (GED) model. The Metropolis–Hastings algorithm is used to estimate the GED model parameters in the Bayesian framework. The application of the GED model is illustrated through the analysis of mathematical development data. Results show that the GED model can correctly identify the deviation from normal of the error distributions. |
| |
Keywords: | growth curve modeling generalized error distribution Bayesian estimation longitudinal data analysis robust statistics |
|
|