Conference Program Home
  My Program

All Times EDT

Abstract Details

Activity Number: 273 - Alignment, Accuracy, Precision: Comparing and Combining Data from Multiple Sources
Type: Topic Contributed
Date/Time: Tuesday, August 9, 2022 : 10:30 AM to 12:20 PM
Sponsor: Survey Research Methods Section
Abstract #322406
Title: Bayesian Hierarchical Model for Combining Probability and Nonprobability Samples Under Unknown Overlaps
Author(s): Terrance D Savitsky* and Matthew R Williams and Julie Gershunskaya and Beresovsky Vladislav
Companies: U.S. Bureau of Labor Statistics and RTI International and U.S. Bureau of Labor Statistics and NCHS
Keywords: Survey sampling; Inclusion probabilities; Bayesian hierarchical models; Nonrandom sample

Estimation of a population quantity derived from a convenience sample will typically result in bias since the distribution of variables of interest in the convenience sample is different from that for the population. A popular set of approaches estimates inclusion probabilities for convenience sample units to allow their combination with the randomly-drawn reference sample under restrictive assumptions. This paper introduces a novel Bayesian hierarchical formulation that simultaneously estimates sample propensity scores and the sample inclusion probabilities from the population for the convenience sample units by conditioning on shared predictors between the two samples and their sample inclusion indicators. Our method jointly models the known inclusion probabilities for the reference sample units along with the sample inclusion indicators. We simultaneously estimate the convenience and reference sample inclusion probabilities for the convenience units. We compare two different methods for forming the sample propensities.

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

Back to the full JSM 2022 program