A0958
Title: Incorporating heterogeneous types of uncertainty in small area estimates from multiple demographic data sources
Authors: Emily Peterson - Emory University (United States) [presenting]
Lance Waller - Emory University (United States)
Abstract: Hierarchical models for small-area estimation have a rich history within the statistical toolbox for analyzing national census data and demographic projections. With the advent of national-level surveys with regional components (e.g., the US American Community Survey, Census data, Demographic and Health Surveys) and data science-based estimates (e.g., WorldPop, Global Burden of Disease, Meta, and Google), there is an opportunity to incorporate multiple heterogeneous data sources to improve the accuracy of local small area estimates. While a hierarchical modeling framework often provides an approach for linking multiple types and layers of population data, there are important contrasts in data collection, data availability, and processing methodologies across data sources, such that each set of population counts may be subject to different sources and magnitudes of error. Firstly, a brief outline of types of US-based small area population estimates and associated errors is provided. Secondly, approaches to robustly fuse information are explored from multiple data sources to improve the accuracy of small area estimates and obtain associated uncertainties in order to provide an inferential framework for such estimates.
