Activity Number:
|
252
- SPEED:Improving Survey Data Quality with Multiple Data Sources, Administrative Data, and Nonresponse Bias Control, Part 2
|
Type:
|
Contributed
|
Date/Time:
|
Monday, July 29, 2019 : 2:00 PM to 2:45 PM
|
Sponsor:
|
Survey Research Methods Section
|
Abstract #307635
|
|
Title:
|
Exploring Hybrid Methods for Estimation with Combined Probability and Nonprobability Samples
|
Author(s):
|
Qiao Ma* and Edward Mulrow
|
Companies:
|
NORC at University of Chicago and NORC at the University of Chicago
|
Keywords:
|
statistical matching;
nonprobability sample;
propensity modeling;
super population modeling;
weighting;
coverage error
|
Abstract:
|
Numerous methods have been developed to combine probability and nonprobability samples via calibration, statistical matching, superpopulation modeling, and propensity modeling. Yang et al (2017) compared and contrasted these methods, and noted that when using statistical matching there is a proportion of probability sample units that do not match to nonprobability sample units. A conjecture based on this observation is that this unmatched portion of the probability sample provides a means to assess the coverage bias of the nonprobability sample. It may also provide a means to account for the bias in estimates based on the combined samples. We propose to explore this conjecture by using statistical matching in combination with either propensity modeling or superpopulation modeling to produce estimates from combined probability and nonprobability samples. Simulations will be used to assess the hybrid methods and determine if estimation bias is reduced versus using propensity or superpopulation modeling alone. We will also empirically examine the methods using a test set of data from a food allergy study in which probability and nonprobability samples were employed.
|
Authors who are presenting talks have a * after their name.