Maximum Likelihood Estimation of Non-Normal Random Effects and Random Errors in Nonlinear Random Effects Models

• In Event: AERA Roundtable Session Saturday 11:40 am Four Seasons Ballroom 4
  In Roundtable Session: Complexity in Data: Simulations and Applications in Complex, Multivariate, Multilevel, and Nonnormal Models (Table 2)

Abstract

Nonlinear mixed effects models (NLMEMs) are widely utilized for fitting repeated measurement (longitudinal) data in educational and psychological research. The basic framework assumes random effect(s) to follow a normal (Gaussian) distribution. Nonetheless, this normality assumption is not always appropriate in contexts where the random effect of a parameter is not normally distributed (e.g., data on reaction time is often positively skewed). The aim of this paper is to demonstrate the maximum likelihood estimation of NLMEMs with non-normal random effects distribution. The study is motivated by a real data example, and it will entail a comprehensive Monte Carlo simulation study to exhibit the robustness, accuracy, and precision of the described estimation algorithm.
Keywords: mixed effects model, nonlinear, non-normal random effects

Authors

• Yue Zhao, University of Minnesota

• Nidhi Kohli, University of Minnesota