Spatial Heterogeneity of Socioeconomic Data: Multiscale Approach and Generalization

The paper discusses the features of applying multiscale approach in studies of spatial heterogeneity. We analyze socioeconomic indicators for different scales of spatial organization in Russia: municipalities, regions (federal subjects), and economic areas (‘economicheskiy rayon’). It is established that more discrete levels of subdivisions, in accordance with statistics theory, have higher levels of heterogeneity. Based on our calculations, we demonstrate that the evaluation of heterogeneity indices for the same territories when applying different grids is combined with partial distortion of these indices. Such errors are explained by the continuity of the geographical space and the failure tos unambiguously determine the true geographical boundaries. We propose a generalization coefficient, a proportion of heterogeneity indices at different scale levels, which enables to assess multiscale spatial heterogeneity. The coefficient provides a means of distinguishing scale levels with the greatest diversity of territories. In addition, this coefficient can be used to differentiate between ‘statistical’ heterogeneity, which is explained by the number of elements in a system, and ‘actual’ (geographical) heterogeneity. A case study of estimating heterogeneity for E.E. Leizerovich’s microzoning grid revealed that the ‘actual’ heterogeneity level can be significantly lower than the one, evaluated using traditional calculations. We provide examples of practical use for the coefficient, e.g., it enables to assess the validity of typologies and elaborate more detailed projections for discrete grids (so called ‘small areas’).