Online Program Home
  My Program

Abstract Details

Activity Number: 48 - Bayesian Adaptive Survey Designs
Type: Invited
Date/Time: Sunday, July 30, 2017 : 4:00 PM to 5:50 PM
Sponsor: Survey Research Methods Section
Abstract #322151
Title: Modeling Nonresponse Bias Likelihood and Response Propensity: The Design and Implementation of Statistical Models to Identify Cases for Interventions During Data Collection
Author(s): Daniel Pratt* and Jeffrey Rosen and Michael Duprey and Jamie Wescott
Companies: RTI International and RTI International and RTI International and RTI International
Keywords: responsive design ; adaptive design ; modeling ; data collection methods ; nonresponse bias
Abstract:

Longitudinal studies benefit from prior information to inform data collection strategies. The presentation describes two models used together during data collection of a US/ED National Center for Education Statistics study to identify cases for interventions. The presentation describes a response likelihood model used to identify, in advance of data collection, likelihood of cases to participate. Using prior data/paradata, we fit a model predicting prior-round response. We used coefficients associated with predictors to estimate response likelihood. The response likelihood model informed decisions about inclusion/exclusion of cases for interventions to control costs. The presentation describes a bias likelihood model used to select cases for interventions. The bias likelihood model was used to identify cases most unlike cases that had already responded at the time the model was run. The model used key survey and frame variables as predictors to identify nonrespondents most likely to cause bias in key survey variables if they did not respond. The model was run multiple times during data collection to identify cases for various interventions (e.g., incentives; field data collection).


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

Back to the full JSM 2017 program

 
 
Copyright © American Statistical Association