Online Program Home
  My Program

Abstract Details

Activity Number: 516 - Estimation with Complex Samples
Type: Contributed
Date/Time: Wednesday, August 2, 2017 : 10:30 AM to 12:20 PM
Sponsor: Survey Research Methods Section
Abstract #324364
Title: Bootstrap Confidence Interval Bands for Estimates in Measurement Error Model with Linked NHIS and EPA Data
Author(s): Rong Wei* and Pavlina Rumcheva and Van Parsons and Jennifer Parker and Ambarish Vaidyanathan
Companies: NCHS / CDC and NCHS/CDC and National Center for Health Statistics and National Center for Health Statistics and NCEH/CDC
Keywords: bootstrap confidence interval ; measurement errors ; health status ; air quality system (AQS) ; linked complex survey data ; heteroscedastic error
Abstract:

The CDC conducts the National Health Interview Survey (NHIS) and maintains a database of air quality estimates from USEPA's Bayesian space-time Downscaler (DS) fusion model, which combines output from the Community Multi-scale Air Quality model with monitor-based measurements from USEPA's Air Quality System. Specifically, DS-based annual predictions and estimated standard errors of 24-hour average PM2.5 concentrations (µg/m3) for 2010 U.S. Census tract centroid locations were linked to NHIS data. Estimated standard errors varied across locations, depending on distances to air monitors. Nonparametric regression with measurement errors was used to associate population health with annual air quality estimates. A deconvolution estimator with heteroscedastic error in R package "Decon" was used. Bootstrap confidence interval (CI) bands for "Decon" results were generated using R package "Boot". Several alternative methods to compute CI bands were applied and compared. Point and CI estimates across methods were compared for models with and without measurement errors.


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

Back to the full JSM 2017 program

 
 
Copyright © American Statistical Association