Online Program Home
  My Program

Abstract Details

Activity Number: 627 - Estimation with Nonprobability Samples
Type: Contributed
Date/Time: Thursday, August 3, 2017 : 8:30 AM to 10:20 AM
Sponsor: Survey Research Methods Section
Abstract #324969 View Presentation
Title: Blending of Probability and Convenience Samples as Applied to a Survey of Military Caregivers
Author(s): Michael Robbins*
Companies: RAND Corporation
Keywords: Inverse probability weighting ; Calibration ; Propensity scores ; Military caregivers
Abstract:

Probability samples are the preferred method for providing inferences that generalize to a larger population. When a small subpopulation is of interest, this approach is unlikely to yield a sample size large enough to produce precise inferences. Convenience sampling may provide the necessary sample size, but selection bias may compromise the generalizability of results. We propose methods of statistical weighting that may be used to combine data from a probability sample with data from a convenience sample. Methods are proposed which involve direct estimation of inclusion probabilities for both samples via propensity scores and which involve calibration weighting. We propose a distinction between methods that blend the samples simultaneously and methods that blend each sample separately. Simultaneous blending invokes a smaller design effect; a test for the adequacy of blending is proposed that uses weights found via disjoint blending. Motivating the exposition is a survey of military caregivers. Simulations show that the gain in precision provided by the convenience sample is lower in circumstances where the outcome is strongly related to the auxiliary variables used for blending


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

Back to the full JSM 2017 program

 
 
Copyright © American Statistical Association