The term census usually refers to the complete process of preparation, collection, compilation, evaluation, analysis, and dissemination of data on demographic, social, economic, and housing characteristics. In most countries, population censuses provide the only complete set of statistical information for all residents. Such information is fundamental for any GIS developed for socioeconomic and, often, environmental analysis purposes.
While surveys collect more detailed information on topics of special interest, they tend to be based on fairly small samples that yield representative information only for large groups of the population. Censuses, on the other hand, collect a small set of essential indicators that are a critical component of the national statistical system. Statistical offices also conduct censuses of agriculture or of the economy (e.g., industrial establishments). In some instances, these are in fact large sample surveys that do not provide complete coverage. This entry focuses on population and housing censuses, which are usually carried out jointly.
The population and housing census is usually conducted by a national statistical agency. In the 2000 census round, censuses were conducted by 190 of the 233 countries and areas for which the United Nations Statistics Division tracks information. These censuses have four main characteristics:
- Universality: The census should cover the territory of the entire country, which ensures that all persons and all housing units are included. Sometimes, this goal is not achievable, for instance, if parts of a country experience conflict or if there are special population groups that are hard to enumerate, such as nomadic people. In those cases, a census may be complemented by statistical sampling.
- Individual enumeration: A census requires that each person or dwelling unit is enumerated individually so that their characteristics are captured separately. This facilitates cross-tabulation of census indicators and compilation of information for small areas.
- Simultaneity: A census is representative of the status at a specified time, usually a single day or a period of a few days in which the information is collected. Some questions, such as the birth or death of family members, refer to a longer reference period, such as the year before the census. A census is defined as de facto if it records persons according to their actual location at the time of enumeration or de jure if persons are referenced at their usual place of residence. This distinction can make a significant difference, for example, in countries with many temporary migrant workers.
- Defined periodicity: An important purpose of a census program is to provide information about changes over time. For instance, demographic projections tend to rely on information from two or more censuses. A census should therefore be conducted at regular intervals. Because of the large cost involved, most countries conduct censuses only every 10 years, although the periodicity is less regular in many developing countries.
Specific questions included in the census will vary from country to country, depending on the level of development, available resources, and special circumstances that determine the need for comprehensive information. For example, censuses in developing countries are an essential source of information on basic services, such as toilet facilities, water supply, or electricity. A number of countries collect information on income, even though the reliability of responses is often questioned. Some information may also be considered essential by some countries, but too sensitive by others—examples are questions on race or ethnicity.
The basic information, however, tends to be quite similar across countries. It includes basic demographic characteristics (age, sex, marital status), migration, fertility and mortality, education, and economic activity. The housing portion of the census will collect information on building types; year of construction; building materials; household amenities, such as bathroom and cooking facilities; access to electricity and water; and tenure arrangements. The information collected for individual household members is linked to the housing information so that social and housing characteristics can be cross-referenced. The United Nations Statistics Division has developed guidelines on census topics that are revisited at regular intervals. In the 2000 round of censuses, for example, countries were encouraged to collect information on disability.
How Census Data Are Collected
The key characteristic of a census is that information is collected for each person and each dwelling unit individually. A census is therefore a massive undertaking even in relatively small countries. Preparations start several years before the census date. Actual information collection is traditionally done by sending individual enumerators door-to-door, each within a specified area of a city or rural region. These enumeration areas (EAs) usually contain as many households as can be canvassed by one enumerator within the period of the census.
A key component of the census process is census cartography. Maps are used in census preparation, during the census, and for postcensus quality control and dissemination. Most important, maps help ensure that every part of the country is included and every person is enumerated, provide a reference for enumerators in the field, and support data management and production of cartographic outputs such as census atlases.
Earlier censuses used hand-drawn EA maps that provided the main physical features delineating EA boundaries as well as other reference points. Many recent censuses have instead used digital mapping, which facilitates updating, reproduction, and use for purposes other than the census. The number of maps required is very large. The 2001 census for the United Kingdom, for instance, required approximately 70,000 maps of EAs and another 2,000 maps of larger census districts, which are collections of individual EAs. In countries where no large-scale, up-to-date base maps are available, creation of a digital map base for census purposes is a time consuming and expensive task. Remote-sensing products and field data collection using global positioning system (GPS) receivers are usually employed in this process. Timor-Leste, for instance, completed a census in 2004 in which each household was digitally georeferenced.
The complexities of enumeration and the enormous data volumes involved in censuses mean that postprocessing and tabulation can take a long time. Delays of several years between the census date and publication of results have been frequent in the past. More recent censuses have employed scanning and intelligent character recognition techniques rather than manual data entry. Some countries have explored use of computer-assisted interviewing techniques over the phone or using handheld computers. In the future, censuses may be conducted over the Internet, as has already been the case on a partial basis in Switzerland, the United States, and Singapore.
Alternatives to a Census
While some countries, such as the United States, have constitutional requirements to conduct a census every 10 years, others, such as Germany and Denmark, have not carried out a census enumeration for several decades. Reasons include high cost, concerns about data confidentiality, and the availability of alternative ways to collect relevant information. Many European countries, for example, have civil registration systems. Residents are required to register with local authorities when they move and report major life events, such as births and deaths. In combination with other local administrative data, this provides essential demographic and social information similar to that from a census. More comprehensive information is then collected using large sample surveys that provide Census———31 statistically accurate information for aggregate entities such as cities or districts.
Individual census records contain personal information that is usually protected by a national statistical law. Individual-level data are therefore not disseminated outside the census office. For publication of reports and census databases, the individual-level information is aggregated to a level that ensures confidentiality. For research purposes, some countries provide anonymized sample data, for instance, a 1% sample of census records from which personal identification such as names and addresses have been removed. In some instances, census offices provide researchers highly controlled access to the unit-record data within the confines of the census office.
Dissemination and Uses
Although censuses are used for macroanalysis of population, social, and housing dynamics, the main benefit of a census is as a source of small-area statistics. Virtually any government planning function and many private sector applications benefit from reliable information on population and social characteristics that is available for small reporting units. Small-area data become even more important as many countries are decentralizing government functions to provinces and districts. In developing countries, this is also reflected in the rapid adoption of poverty mapping, a small-area estimation technique that generates high-resolution information on welfare-related indicators by combining census micro data with detailed household survey information. Census cartography also often plays an essential role in a country’s national spatial data infrastructure. The statistical office tends to be the custodian of administrative reference maps that aggregate up from EAs to subdistricts or wards, districts, and provinces. These boundaries also serve as reference for many other socioeconomic data and are therefore a key component of a national GIS framework database.