|
Activity Number:
|
133
- Statistical Issues in Environmental Epidemiology and Pharmacoepidemiology
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 9, 2021 : 1:30 PM to 3:20 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
|
Abstract #317918
|
|
|
Title:
|
Integrating Conflicting Exposures from Secondary Commercial Data Sources
|
|
Author(s):
|
Jung Yeon Won* and Brisa Sanchez and Michael R. Elliott
|
|
Companies:
|
University of Michigan and Drexel University and University of Michigan
|
|
Keywords:
|
Data Integration;
Built-environment;
Measurement Error;
Secondary data;
Commercial business list;
Count exposure
|
|
Abstract:
|
One challenge in built-environment research is ascertainment error that arises in mering multi-source information. Secondary commercial lists often provide conflicting claims about the presence and location of businesses. Consequently, disagreement among exposures collected from different sources gives biased inferences. Our work shows that bias in the naïve regression model depends on the dispersion of true exposure and the source credibility. Instead of constructing a parametric model for measurement error, we employ the knowledge of the reliability degree of databases from field validation studies to effectively integrate exposures from different sources. We can derive complete information about latent true exposures using incomplete data tables with partially known margins. Our method uses a Bayesian nonparametric model to make few distributional assumptions about the counts of businesses in a region and handle the overdispersion in exposures. We demonstrate our model's utility in ecological analyses of finding an association between school-level obesity and the surrounding food environment.
|
Authors who are presenting talks have a * after their name.
