Activity Number:
|
157
- Constructing Profiles of Local Communities
|
Type:
|
Topic Contributed
|
Date/Time:
|
Monday, July 30, 2018 : 10:30 AM to 12:20 PM
|
Sponsor:
|
Social Statistics Section
|
Abstract #330348
|
Presentation
|
Title:
|
Constructing a Synthetic Population for Community Profiling Using Publicly Available Data
|
Author(s):
|
Joshua Goldstein* and David Higdon
|
Companies:
|
Social and Decision Analytics Laboratory, Virginia Tech and Virginia Tech
|
Keywords:
|
synthetic population;
community profiling;
multiple imputation;
Markov chain Monte Carlo
|
Abstract:
|
We demonstrate how one can use information from publicly available data sources to construct a realistic synthetic population. As a demonstration we focus on two counties in northern Virginia. Some features of the synthetic population are an exact match to corresponding features of these counties; others are generated statistically, using survey and administrative data collected at various levels of aggregation. Methodology used to construct this synthetic population combines ideas from traditional methods for generating synthetic populations with concepts from multiple imputation and Markov chain Monte Carlo. This results in an ensemble of synthetic populations that can then be used to make inferences at resolutions not available in any single data source alone. We consider two examples of this methodology; One in assessing uniqueness of public use files released by the American Housing Survey in Arlington County, and the other in estimating factors that influence obesity in Fairfax County.
|
Authors who are presenting talks have a * after their name.