Activity Number:
|
139
- Improving Population Inference Using Statistical Data Integration
|
Type:
|
Topic Contributed
|
Date/Time:
|
Monday, August 8, 2022 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Survey Research Methods Section
|
Abstract #322645
|
|
Title:
|
Analysis of Data Combined from Multiple Sources in the Presence of Linkage Error
|
Author(s):
|
Martin Slawski* and Brady Thomas West and Emanuel Ben-David
|
Companies:
|
George Mason University and Institute for Social Research, University of Michigan-Ann Arbor and United States Census Bureau
|
Keywords:
|
Record Linkage;
Mismatch error;
Pseudo-Likelihood ;
Small Area Estimation
|
Abstract:
|
Record linkage bears considerable promise for the generation of rich data sets from existing sources, often at no additional cost. However, record linkage is rarely free of error: missed matches (matching records not identified as such) and false matches (records belonging to different entities) can substantially contaminate statistical analyses performed on linked data sets. We discuss the impact of such errors in different scenarios and provide an overview of several mitigation strategies and the associated computational implementations. Specific emphasis will be given to situations with no or only very limited information on the linkage process. Finally, we study the application of those strategies to small area estimation based on linked covariates, a problem particularly relevant to finite population inference.
|
Authors who are presenting talks have a * after their name.