Census and sensibility

Stan Openshaw's invaluable handbook is a guide to the contents of the census, the statistical packages and analytical methods for gaining access to small area data and worked examples. It is, however, more than a users' manual: it is a debate about, and in parts a trenchant critique of, a system that is increasingly coming under strain due to rapid technological advances and the much wider availability of alternative sources of data.

Although the census relates to individuals and households, this information is aggregated to geographical areas, and users of the census data have no access to individual records. The result is that, in order to meet a wide diversity of information needs, the Office of Population, Censuses and Surveys (and its counterparts in Scotland and Northern Ireland) produced in 1991 a large number of output tabulations. In addition to the complexity that resulted, the overriding criteria of anonymity constrained the way the information of most interest - that on subnational areas provided by the small area statistics and local base statistics - could be produced. But 1991 also introduced the samples of anonymised records, of 1 per cent and 2 per cent samples of data for anonymous households and people respectively: a milestone in the development of household survey analysis.

The book highlights four major methodological criticisms: 1. The rapid obsolescence of data; 2.The 1991 census's "missing million" - an underenumeration that is particularly large for men aged 2029 living in metropolitan areas; 3.The use of enumeration districts as the basic building blocks for data collection - these suffer from problems of comparability across local authority areas and between censuses. The Scottish solution, of using output areas - in effect, postcodes - is urgently needed in England and Wales; 4.The relationship between the census and geographic information system statistical packages. Essentially, GIS provides a comprehensive set of mapping tools that facilitates easy linkage with other spatial information to provide multi-source datasets. These GIS techniques are not, however, being incorporated into the management of the census system quickly enough: in effect, the British system is divided between the digital map data owned by the Ordnance Survey and the related attribute data owned by various government departments.

Openshaw's final chapter argues that the census probably has no future in its present form. The main problem with current information processing is the lack of integration of the various data sources available. Since this handbook was published, the government has announced the merger of the Government Statistical Service and the OPCS into a new Office for National Statistics, which should go some way to address Openshaw's criticism. But this merger will still leave the national mapping agency (that is, the Ordnance Survey) outside the frame.

Daniel Dorling's atlas shows what can be achieved to provide information about social structure in visual form, using a range of census materials and other data bases. The maps are population cartograms; that is, combined maps and graphs that allocate equal areas of the map to equal numbers of people. Each ward is represented by a circle, with the circle's area proportional to the population of that ward. Colour coding shows the gradations of the social patterns. The atlas divides the cartograms and accompanying analysis into chapters on population, demography, economic configuration, housing, health, society, and politics. The persistence of the inequalities in Britain is made vividly immediate through these visual techniques. Work patterns not only diverge between north and south but between younger and older members of the workforce and between the sexes. Since 1981 men in full-time work, particularly those over 45, have suffered the largest job losses, notably in coal mining and manufacturing, while employment in service industries has grown. But it is young adults, and those from ethnic minorities, who are most likely to be unemployed. Overall, by 1993 male unemployment in Britain was higher than at any point since the 1930s.

Housing and health show similar divisions. In the 1990s all the traditional housing problems - space, quality, and affordability - remain, together with the new ones of mortgage and rent arrears. In health, the highest death rates from heart, circulatory and respiratory diseases, and from cancer and suicide, are to be found in the north and in cities. This spatial pattern has remained relatively unaltered over at least the past 40 years; if anything, the mortality rate map is reverting to the picture before the second world war. Demographically, there is a decline in marriage rates, and an increase in divorce rates, which reflect fundamental changes in family structures. The type of family that declined most rapidly over the 1980s was the "traditional family" of two adults, one of whom was in work, and two children. By 1991, one child in ten lived in a household in which no adult was in paid employment, and these households are increasingly concentrated in the north and in Wales. Since 1979, moreover, the net income of the richest fifth of households has risen to more than seven times that of the poorest fifth; these divisions are manifested in spatial terms, with the rich and poor increasingly living apart. Dorling's visual tour-de-force clearly demonstrates that geographically, class differences underlie most patterns of lifestyles and life chances.

Dilys M. Hill is reader in politics, University of Southampton.