A Block-level Estimation of Residential Characteristics Using Survey and Spatial Microdata

Abstract

A wide variety of microdata offer great potential for the advancement of small area demographic estimation. This study proposes a method to estimate residents’ attributes at the street block level using survey microdata and spatial microdata (GIS building data), demonstrating the method in a Tokyo ward area. The proposed method is shown to be superior to alternatives using limited microdata. This method includes (1) modeling housing composition for each block based on dwelling unit information from spatial microdata, (2) establishing the statistical relationship between housing and demographic attributes through a small area cross tabulation technique using survey microdata, and (3) allocating demographic attributes to street blocks based on this small area-specific relationship. Housing composition is defined by building type and dwelling size and is constrained by small area totals of the census. The Tokyo ward area contains high-density built-up districts and is characterized by vertical population extension and resident diversity. In spite of this complex residential structure, the accuracy assessment indicates that the proposed method performs well and presents a more accurate estimation than any other methods using limited microdata. This finding stresses the effectiveness of combining different types of microdata, which not only allows spatial refinement but also reflects neighborhood heterogeneity of the relationship between housing and demography.