census

census A complete and individual enumeration of all cases of the type specified within defined boundaries at a single point in time; a 100 per cent count of some social entity or type of event. In contrast to a survey, in which only part of a population is included, a census yields cross-sectional data, ranging from a simple head-count to more sophisticated information, on every member (in theory at least) of a population.

In order to achieve such complete coverage, national censuses usually require compulsion, an obligation to participate and co-operate in providing the information required, and are therefore the preserve of national governments. National censuses have been undertaken only relatively recently (the first British Census took place in 1801).

Most countries carry out a population and housing census every ten years with compulsory participation to ensure total coverage and a complete head-count. Population data in inter-censal years must then be inferred from inter-censal projections or from sample surveys. Some countries also have other national censuses, for example of employment or business activities, or industrial output. Censuses may be carried out annually, or at periodic intervals, instead of every ten years—although there are obvious constraints of logistics and finance. In recent years, therefore, there have been experiments in applying sampling techniques to national censuses. What this usually means is that the 100 per cent count is limited to identifying all relevant cases, combined with a questionnaire survey of a specified proportion of all cases, ranging from 10 per cent to 50 per cent—in effect a census followed immediately by a sample survey. Other bodies that attempt censuses—for example of all members of an association—may achieve high levels of co-operation, but are unlikely to achieve complete coverage due to the absence of any compulsion to participate, resulting in some degree of non-response.