Spatial scale and small area population statistics for England and Wales

It is well-known that the results of analyses of aggregate data, such as those provided as outputs from censuses, are dependent on the size and shape of the zones used to report the data. However, many users of aggregate census data do not consider how far the zones utilised in their analyses capture spatial information about the population sub-groups they are studying. In addition, future data collection strategies should account for such issues. This article takes as its focus England and Wales, and it seeks to assess how far output areas (OAs) and aggregations of OAs capture information in selected population sub-groups and, therefore, how important it might be to use zones of a particular size in order to properly analyse the geographies of these subgroups. The article uses the index of dissimilarity, Dxy (for groups x and y), and variograms to assess spatial variation in population sub-groups as represented by counts for OAs, lower layer super output areas (LSOAs), middle layer super output areas (MSOAs) and local authority districts (LAs), all produced as outputs from the Census in England and Wales. The analyses show how much information is contained at each spatial scale for sub-categories relating to age, ethnic group, housing tenure, car or van availability, qualifications, employment, limiting long-term illness (LLTI) and National Statistics Socio-economic Classification (NS-SeC). The amount of variation contained by each level of the hierarchy of zones differs markedly by population sub-group. LAs capture most (83%) of the variation in the spread of the population by the binary categorisation of White/Black, Asian and Minority Ethnic (far more than for any other variable), but there remains considerable local variation. The results suggest that zones larger than OAs are not geographically detailed enough to enable meaningful analysis of local-level differences between places and thus any alternative to the Census in the United Kingdom (with England and Wales as a specific case) must provide zones equivalent in size to OAs. If estimates are available only for larger areas then much information will be lost and our ability to explore how sub-group characteristics, or the relationships between variables, differ between localities will be considerably diminished. The results also provide evidence on some of the ways in which the population of England and Wales was geographically distributed in 2011. ARTICLE HISTORY Received 17 March 2015 Accepted 18 October 2015