Exploring the Modifiable Areal Unit Problem

• Activity Number: 488
• Type: Contributed
• Date/Time: Wednesday, August 12, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Government Statistics Section
• Author(s): Talha Ali* and Owais Gilani
• Companies: Yale School of Public Health and University of Michigan School of Public Health
• Keywords: spatial epidemiology ; areal unit problem ; spatial clustering ; spatial resolution ; household income disparity ; US Census

Abstract

The modifiable areal unit problem refers to introduction of statistical bias in hypothesis testing when spatial processes are aggregated over arbitrary areal units. Aggregation of a spatial process over different areal units can result in varying conclusions, so that the results of the analysis are dependent on an arbitrary choice of areal unit. In spatial epidemiology, this problem is of particular concern as the spatial resolution of disease incidence available for analysis is generally limited due to privacy concerns. Our goal is to explore the extent of the modifiable areal unit problem by comparing results of a simple analysis conducted at the census tract and census block level. We explored the spatial trend in aggregate household income (AHI) from US Census Bureau's 2014 Planning Database, and performed a cluster analysis to identify areas of spatial clusters of low aggregate household incomes. We also explored the spatial trend in AHI after adjusting for potentially correlated demographic covariates. Spatial trend of the clusters in both of these analyses is compared across census tracts and census blocks to explore how choice of areal unit of analysis impacts results.