/science /census

census | Facts, Definition, Methods, & History

Census, an enumeration of people, houses, firms, or other important items in a country or region at a particular time. Used alone, the term usually refers to a population census — the type to be described in this article. However, many countries take censuses of housing, manufacturing, and agriculture.

Censuses, being expensive, are taken only at infrequent intervals: every 10 years in many countries, every 5 years or at irregular intervals in other countries. In noncensus years the population is estimated with the aid of vital statistics if these are sufficiently reliable. For instance, the population in postcensual year X equals the population at the last census plus the births, minus the deaths, plus or minus the net migration during the intervening years.


Strictly speaking, the modern population census began to evolve only in the 17th century. Before that time, inventories of people, taxpayers, or valuables were certainly made, but the methods and purposes of such inventories were different from modern ones. The most important difference was that early inventories were made to control particular individuals — e.g., to identify who should be taxed, inducted into military service, or forced to work. Because it was usually not to an individual's interest to be counted or to give correct information for these purposes, the premodern enumerations tended to be inaccurate. A second difference was that early inventories did not seek to count all the people or even a representative sample of them but only those in particular categories, such as family heads or males of military age. Such surveys are known to have been made in ancient Babylonia, Palestine, Persia, China, and Egypt. Every five years, the Romans enumerated citizens and their property to determine their liabilities. This practice was extended to include the entire Roman Empire in 5 BC. After the collapse of Rome the practice was discontinued in the West until the modern period. The main exception was Domesday Book, the inquest of England in 1086 that was made to acquaint William the Conqueror with the landholders and holdings of his new domain. Under the threat of siege, the German city of Nürnberg made an almost complete count of its people in 1449.

The modern idea of a population census as a complete enumeration of all the people and their important characteristics for purposes of understanding the basic structure and trends of the society rather than for identifying and controlling particular individuals slowly arose in the 17th and 18th centuries. There is no such thing as "the first census" because, although early efforts embodied one or another modern feature, none of them embodied all modern features. Perhaps the first effort in an area larger than a city to count everyone at successive intervals was made in New France (Quebec) and Acadia (Nova Scotia), where 16 enumerations were made between 1665 and 1754. In 1749 the Swedish clergy, who had long kept lists of parishioners, was required to render returns from which the total population of Sweden (including Finland) was obtained, and a similar development began in Denmark in 1787. Certain Italian states conducted approximately accurate enumerations: Sardinia in 1773 and 1795, Parma in 1770, and Tuscany in 1766. Enumerations occurred in several German states from 1742 onward. Although several British North American colonies — such as Virginia in 1624-25 and subsequently — had made full enumerations, the United States made history when it took its first census in 1790, not only because of the size of the area enumerated and the effort to obtain data on characteristics of the population but also because of the political purpose for which it was undertaken — namely, representation in Congress on the basis of population. England took its first census in 1801, and, although France tried to do so in 1800 and 1806, the administrative machinery was poor and the results untrustworthy until 30 years later. Only gradually did the early census takers learn what was basic information and how to obtain it. The first United States census did not, for example, secure data on occupation, birthplace, marital status, or exact age. A five-year age classification of whites was provided by the 1800 census, but other items had to wait until later, and until 1850 the unit used was the family rather than the individual. In England no satisfactory attempt was made to secure age data until 1841, and no effort to secure marital status until 1851.

The history of modern census taking thus involves three parallel developments: (1) the invention and gradual spread of the idea of countrywide enumerations for general scientific and governmental purposes, (2) the improvement of administrative machinery, techniques, and accuracy of enumeration, including legal safeguards assuring that individual answers would be kept in confidence, and (3) the deepening and systematization of the types of information obtained. All three developments tended to increase the effectiveness of censuses as sources of essential information.

It is a mistake to think that in the history of each country there is a date, marked by the first census, before which the size and characteristics of the population were totally obscure and after which they are totally clear. Usually some fragmentary information was available even before the first census, and afterward some time was required before the census record became reasonably full and accurate. As the countries pioneering in census taking gradually worked out better systems of enumeration, however, these could be transferred to new countries without repeating all the trials and errors. Canada's first dominion-wide census was taken in 1871, India's in 1871-72, and Egypt's in 1897.

In addition to the national censuses, others were taken in colonial territories, parts of countries, cities, and so on. It is estimated that in the decade after World War II at least 150 countries or areas took censuses collecting individual data on more than two billion persons. The large number of countries having censuses in the second half of the 20th century was partly due to efforts of the United Nations. Not only does the United Nations encourage countries to take censuses, but it sponsors regional statistical committees that suggest minimum standards and offer technical assistance in the planning and conduct of enumerations. When China reported a census in 1953, the last large part of the world was removed from demographic darkness. The population of the entire world is now known with a degree of accuracy never before approximated, and the structure of populations (including their socioeconomic characteristics) and their patterns of change are understood as never before.

Modern census procedure

Modern censuses refer to a precisely delimited territory and subareas and, for this reason, are normally planned and conducted with the aid of detailed maps. They aim to enumerate every person within the designated territory. A "de jure" census tallies people according to their regular or legal residence, whereas a "de facto" census allocates them to the place where enumerated — normally where they spend the night of the day enumerated. By either method, the reported territorial distribution is according to where people sleep (nighttime population) rather than where they work (daytime population). For this reason the population of central cities is usually far less than the number employed there and that of suburbs far more than those employed in suburbs.

In addition to being precise as to territory, the census must be precise as to time; accordingly, a specific moment is almost always selected. This "census moment," often fixed at midnight, becomes the chronological line separating the included from the excluded. All persons born after the census moment or dying before it are excluded; all others are included. The census moment is also the reference point for certain kinds of questions, such as age, marital status, and citizenship. Although some questions, such as occupation or labour-force participation, refer to a period of time rather than an infinitesimal moment, clarity requires that the exact limits of the period be specified. The need for time precision is necessary because the actual enumeration may take days or weeks. Because during this period many people will have moved about, died, or been born, the census is taken as of a certain day, hour, and moment.

The requirement that all individuals be enumerated does not mean that all individuals must actually be interviewed. A wife may supply information about her husband; a parent may supply information about his or her children. Nor does it mean that every question must be answered about every individual. If all individuals are enumerated — that is, counted and categorized on a few basic questions — there then exists a basis for sampling with respect to other questions.

Because census information is obtained by using a fixed questionnaire for interviewing, there are two broad types of resulting data: direct data, the answers to specific questions on the schedule; and derived data, the facts discovered by classifying and interrelating the answers to various questions. Direct information, in turn, is of two sorts: items such as name, address, and the like, used primarily to guide the enumeration process itself; and items such as birthplace, marital status, and occupation, used directly for the compilation of census tables. From the second class of direct data, derived information is obtained, such as total population, rural-urban distribution, and family composition. The following is one list of desirable topics suggested by the United Nations.

Direct topics:
Geographic(1) place enumerated and/or place of usual residence
Familial(2) relation to head of household or family
Demographic(3) sex, (4) age, (5) marital status, (6) children ever born, (7) birthplace
Economic(8) type of activity, (9) occupation, (10) industry, (11) employer-employee status
Social and political(12) citizenship, (13) language, (14) ethnic or religious affiliation
Educational(15) literacy or level of education, (16) school attendance.

Derived topics:
(17) total population;
(18) population of towns and local areas;
(19) urban-rural distribution;
(20) household or family composition.

Possible additional topics include prior place of residence (for understanding internal migration), farm-tenure status, income, labour-force participation, duration of marriage, and number of prior marriages.

The machinery used to analyze the results has a profound influence on the whole census. In the 19th century, when tabulations were made manually, no complex questions were asked because of the impossibility of analyzing them. This situation was entirely changed with the introduction of computerized tabulations and analyses. Modern computers have led to a second revolution that by permitting complex tabulations to be made quickly, has moved the bottlenecks to the data-transfer stages, when the written records are converted for computer use and when the tabulated results are to be turned into printed books.

The Hollerith census tabulator

This cover of Scientific American, August 30, 1890, displays various aspects of Herman Hollerith's invention. IBM Archives

** End of page