Activity Number:
|
202
- Monte Carlo Methods and Simulation I
|
Type:
|
Contributed
|
Date/Time:
|
Tuesday, August 4, 2020 : 10:00 AM to 2:00 PM
|
Sponsor:
|
Section on Statistical Computing
|
Abstract #314048
|
|
Title:
|
Longitudinal Treatment Effect and Non-Ignorable Missing Values of a Binary Response
|
Author(s):
|
Nikhil S Padhye* and Diane Santa Maria
|
Companies:
|
University of Texas Health Science Center at Houston and University of Texas Health Science Center at Houston
|
Keywords:
|
Missing not at random;
Randomized controlled trial;
Multiple imputation;
Sensitivity analysis;
Bayesian model;
Binary response
|
Abstract:
|
Longitudinal evaluation of the treatment effect in a randomized controlled trial can often present the problem of intermittent missing values and dropouts that increase over time. This is particularly true for self-reported data that depend on participant initiative, as in ecological momentary assessments (EMA) communicated on a mobile phone platform. In this study, we consider a binary response variable with a pattern of missingness that could depend on the missing values. Multiple imputation and sensitivity analysis with tipping-point visualization shed light on the limits of validity of the treatment effect. Application is made to EMA data for drug use in youth experiencing homelessness. The Bayesian linear mixed effects model accounts for the presence of active drug users and non-users, in addition to the interaction effect of treatment duration and randomized treatment groups. The impact of missingness patterns of drug use on the treatment effect are evaluated with the aforementioned approach.
|
Authors who are presenting talks have a * after their name.