Online Program Home
My Program

Abstract Details

Activity Number: 428 - Contributed Poster Presentations: Health Policy Statistics Section
Type: Contributed
Date/Time: Tuesday, July 31, 2018 : 2:00 PM to 3:50 PM
Sponsor: Health Policy Statistics Section
Abstract #328584
Title: Index of Local Sensitivity to Non-Ignorability for Longitudinal Data with Non-Monotone Missingness
Author(s): Chengbo Yuan* and Donald Hedeker and Robin Mermelstein and Hui Xie
Companies: University of Illinois at Chicago and University of Chicago and University of Illinois at Chicago and SPH,University of Illinois at Chicago and Faculty of Simon Fraser University
Keywords: longitudinal data; non-ignorable missingness; outcome and covariates missing simultaneously; mixed effect model

There is often a need to perform sensitivity analysis to assess the potential impact of non-ignorable missingness. However a sensitivity analysis that directly fits non-ignorable models can be challenging to conduct because the likelihood functions involve high dimensional integrations, which is complicated by missingness in both outcome and covariates as well as non-monotone missingness patterns in longitudinal data. Gao et al. proposed a method using both linear and non-linear index of sensitivity to nonignorability to measure the local sensitivity of the missing at random estimators to nonignorability for cross-sectional data and models. This method largely simplifies the computation process, and allows the outcome and part of the covariates to be missing simultaneously. We extend the method to longitudinal mixed effects models. We considered both outcome and covariate to follow multivariate normal distribution situations and model the non-monotone missingness using transitional multinomial models. We derive the formulas for ISNI index measures and evaluate the performance of the extended method using simulated data, and apply it to a real EMA dataset.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2018 program