Online Program Home
  My Program

Abstract Details

Activity Number: 149 - Advances in Modeling Multilevel Observational Data from Complex Surveys
Type: Topic Contributed
Date/Time: Monday, July 31, 2017 : 10:30 AM to 12:20 PM
Sponsor: Section on Statistics in Epidemiology
Abstract #322876
Title: Modeling Count Data from a Complex Survey using Pseudo Maximum Likelihood
Author(s): Mulugeta Gebregziabher* and Lin Dai
Companies: MUSC and Medical University of South Carolina
Keywords: complex survey ; demographic health survey ; empirical varaince ; pseudo-likelihood ; sampling weight
Abstract:

In this study we introduce a multilevel pseudo maximum likelihood (MPML) method for weighted count data from a complex survey. We conduct a simulation study to assess the finite sample performance of this estimation method under several scaling methods of the survey sampling weights. In addition, different approaches for estimating the variance of parameter estimates corresponding to covariates that are included in the model are considered and compared in the simulation studies. The simulation results show that MPML estimation with scaled sampling weight provide better performance in terms of relative bias and 95% confidence interval coverage compared to estimations by standard approaches that use sampling weight or that ignore sampling weight. Thus, we recommend MPML to analyze multilevel weighted count data. Among the four variance estimation approaches we considered, the empirical variance estimator provides values that reflect the true variance of the estimators. We demonstrate the proposed methods using real data from national surveys of three countries in sub-Saharan Africa.


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

Back to the full JSM 2017 program

 
 
Copyright © American Statistical Association