Speaker: Dr. Emil Coman
Title: Mapping, visualizing and analyzing mechanisms behind social vulnerability and health disparities using spatial data
Abstract:
Spatial data is increasingly becoming broadly available, and with it comes the unrestrained appetite for drawing conclusions about relations between neighborhood indicators. Drawing on a recent warning paper on improperly reporting plain correlations between spatial variables, I present simple visual ways of analyzing such data, using the 100+ years old path analytic method, adapted for spatial data.
One example uses CT census tract level data: % non-White residents, % without highschool, % unemployed, % in poverty, income, and life expectancy, which are examined to uncover the direction of effects between each pair; I then compare the emerging causal/path analytic model to the one proposed by the Social Vulnerability Index (SVI) factor analytic model. The path analytic model shows that residents from CT census tracts with 10%p more minorities (expect to) live shorter lives by 4 months, 54% of it because of lower income, and 46% because lower education.
A second example uses a weight loss intervention dataset from CT (N = 747 residents), who come from 265 of the 827 CT's census tracts (PI Leahey). Simple spatial regressions indicate that census tract income affects the outcome (% weight lost) stronger than individual income, and multilevel path analytic (SEM) models uncover variable effects of engagement with the intervention on % weight lost across CT census tracts, with systematically weaker such effects in higher social vulnerability census tracts (with lower ‘SES’).
Date Time Location
Friday, September 23rd | 12:15 PM - 1:15 PM | AUST 445
For more information, contact: Katie Lamb at katie.lamb@uconn.edu