|
Activity Number:
|
407
- Novel Methods for Causal Inference in Health Policy
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Tuesday, July 30, 2019 : 2:00 PM to 3:50 PM
|
|
Sponsor:
|
Health Policy Statistics Section
|
|
Abstract #306519
|
|
|
Title:
|
Estimation of Average Causal Effect in Clustered Data Using Multiple Imputation
|
|
Author(s):
|
Recai Yucel* and Meng Wu
|
|
Companies:
|
SUNY Albany School of Public Health and Department of Health, NY State
|
|
Keywords:
|
missing data;
multiple imputation;
causal inference;
Rubin causal model
|
|
Abstract:
|
Causal inference can be seen as a missing data problem within the framework of potential outcomes. In this paper, we estimate ACE and its variance in clustered observational data using multiple imputation (MI) paradigm. We propose two approaches for MI inference on ACE. First is based on all observations, and the second method is based on observations from treatment and control group separately. Our approaches consider different circumstances where treatment can be applicable to the entire population or only to the treated group. Markov Chain Monte Carlo(MCMC) techniques under multivariate extension of linear mixed-effect models are used to sample from the predictive distribution of missing data. We then compute the ACE and its variances from the complete data sets and combine them using MI combination rules. We conduct simulation studies to assess the sampling performance of the two imputation approaches and present a data application where we study the average causal effect of inadequate prenatal care on birth weight for low income women in New York State.
|
Authors who are presenting talks have a * after their name.
