Address- versus postcode-based building blocks for the creation of small area statistical geographies: some lessons from the 1911 British Population Census

Postal geography is increasingly used to define spatial units for data collection and statistical reporting in British population censuses. There has been a shift from defining a single set of spatial units for each census to creating areas with greater consistency across censuses using artificial polygons around collections of postcoded addresses. However, the question remains whether creating new zones for the purpose of reporting aggregate statistics can feasibly be carried out by starting from individual addresses rather than postcode units, while at the same time preserving the confidentiality of households and individuals. Although the issues could be approached by using modelled data derived from modern censuses, the approach adopted here has been to use genuine household and individual level data from the 1911 Census for addresses that have been georeferenced in a selection of local authorities. Testing the difference between address- and postcode-based new zones was carried out by aggregating from two sets of Thiessen polygon building blocks, one for individual 1911 census addresses and the other for these addresses grouped within modern unit postcode boundaries. The sets of new zones were assessed by measures of target population attainment, homogeneity and shape compactness and in respect of how they allocated different percentages of the addresses to the same spatial unit. People and households were distributed at lower densities across new zones that occurred in relatively unpopulated areas, although even in more densely inhabited parts of the study areas some people and households were colocated in the same new zones produced from the two starting sets of different Thiessen polygons.