The geostan R package for Bayesian spatial analysis: strategies for modeling and measuring health inequalities
Topics: Medical and Health Geography
, Spatial Analysis & Modeling
,
Keywords: health geography, health inequality, Bayesian inference, spatial statistics, open source
Session Type: Virtual Paper Abstract
Day: Monday
Session Start / End Time: 2/28/2022 09:40 AM (Eastern Time (US & Canada)) - 2/28/2022 11:00 AM (Eastern Time (US & Canada))
Room: Virtual 37
Authors:
Connor Donegan, The University of Texas at Dallas & UTSouthwestern Medical Center
,
,
,
,
,
,
,
,
,
Abstract
The geostan R package supports spatial analysis workflows for public health research, and facilitates the application of Bayesian inference to research on social and spatial disease distributions. As an accessible interface to the Stan modeling language for Bayesian inference, geostan encourages users to work with full probability distributions, rather than point estimates, while conceptualizing probability as an evolving relationship between theory and evidence. A core motivation for the design of geostan is to accelerate the development of new standards and workflows for analyses that utilize small-area survey data. Methodologies that enable the incorporation of data reliability information into study designs, analyses, and peer reviews, are important because the sampling error in survey estimates, such as those produced by the American Community Survey (ACS), is often substantial in magnitude and socially patterned. A spatial analysis may utilize dozens, hundreds, or even thousands of survey estimates, amplifying the need for modeling strategies that address measurement error with spatial data. Modeling survey-based covariates jointly with health outcomes in geostan can result in inferential models that better reflect the available evidence. A byproduct of building models for covariates themselves is that researchers may be confronted by lurking challenges of study design---such as trade-offs between data quality and conceptual cogency in choice of covariates---and by the practical limitations of spatial data. This presentation will explore these ideas while demonstrating a geostan workflow for measuring socio-spatial health inequalities using U.S. county mid-life mortality data and a new race-class Index of Concentration at the Extremes (ICE).
The geostan R package for Bayesian spatial analysis: strategies for modeling and measuring health inequalities
Category
Virtual Paper Abstract
Description
This abstract is part of a session. Click here to view the session.
| Slides