The Modifiable Areal Unit Problem in the Context of a Standard Grid

Abstract

Data associated to territorial units, such as socio-economic, demographic or epidemiological data, can lead to different conclusions if the territorial units change. For example the correlation between variables often increases when data are aggregated to larger units. Transforming data from an areal system to a different one often involves two steps: disaggregation into small units and re-aggregation. Disaggregation is always possible, but generally introduces a considerable inaccuracy. The quality of disaggregation depends above all on the quality of covariables used to guide the attribution of values to different units. If no covariables can be used, simple areal weighting is possible, but this option is not recommended in general.