A Shared Parameter Location Scale Mixed Effect Model for EMA Data Subject to Informative Missing

• Activity Number: 582 - Random Effects and Mixed Models
• Type: Contributed
• Date/Time: Wednesday, August 1, 2018 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Xiaolei Lin* and Robin Mermelstein and Donald Hedeker
• Companies: The University of Chicago and University of Illinois at Chicago and University of Chicago
• Keywords: Mixed Effects Model; Missing Data; Intensive Longitudinal Data

Abstract

In this paper, we address the problem of informative missing in the context of Ecological Momentary Assessment studies, where each study unit gets measured intensively over time and intermittent missing is usually present. We present a shared parameter model approach that links the primary longitudinal outcome with potentially informative missing by a common set of random effects that summarizes subjects' specific traits in terms of their mean (location) and variability (scale). The primary outcome, conditional on the random effects, is allowed to exhibit heterogeneity with respect to both the mean and within-subject variance, and are further assumed to be independent of the missing process. Unlike previous methods which largely rely on numerical integration or approximation, we estimate the model by a full Bayesian approach using MCMC. An adolescent mood study example is illustrated together with a series of simulation studies. Results in comparison to the naive methods suggest that the proposed model can dramatically increase the model fit yet provide a deeper understanding of how missingness can affect the inference for the primary outcome.