In this paper we use Voronoi polygons (VP) to test weather more homogenous exposure areas can be generated to reduce "ecological bias". In the analysis, soil contamination measurements by Lead (Pb) were superimposed upon a layer of small census areas (SCA) and the average exposure fir each SCA was calculated. Next, VPs were formed around soil test points, with each polygon containing exactly one Pb soil measurement. Spatial interpolations were also run, to compare their results with the results obtained by SCA averaging and VP rezoning. Next, OLS and Spatial Lag regressions were run to link Pb exposure with the health status of local children, with health information retrieved from the Clalit Health Services' database. Model fits were consistently higher in the VP models compared to the SCA and interpolation models, indicating that the VP method appeared to improve the models' explanatory power by reducing exposure misclassification.
|Number of pages||16|
|Journal||Geography Research Forum|
|State||Published - 1 Dec 2012|
- Ecological bias
- Exposure misclassification
- Soil contamination
- Voronoi polygons