Census Implementation

History of Census Taking

It is not known when the first census was taken, but there is evidence of census taking nearly 6000 years ago, with occurrences recorded in Egypt, China, Babylonia, Palestine, and Rome. Early censuses were undertaken especially for the purpose of military conscription, but were also used for purposes of taxation and labor requirements of the state. Thus, in most cases, enumeration only involved adult male citizens.

A modern census is one which meets standards of completeness, accuracy, and simultaneity (discussed further below). These censuses also have broader purposes than the earlier censuses, and these include government planning for provision of services and often the determination of electoral boundaries. The first modern census is sometimes attributed to Sweden in 1749, but there were earlier censuses with modern attributes undertaken in Nouvelle France (Quebec), Acadia (Nova Scotia), and some of the Italian principalities. In the late eighteenth and through the nineteenth-century, many other countries initiated full censuses: United States 1790, Spain 1798, England 1801, France 1801, Ireland 1811, Norway 1815, Australia 1828, Greece 1836, New Zealand 1851, Switzerland 1860, Italy 1861, and Canada 1871.

Through the twentieth century some colonial powers initiated censuses in their territories, although in some cases full censuses were not undertaken until independence. By the end of the century, national censuses had become the norm for nation states. In the 2000 round of census taking (1995–2004), about 200 nations undertook at least one census. Those that did not undertake a census tended to be nations with serious political problems such as civil war.

A new trend in recent years should be noted. Sweden has not undertaken a census after 1990, and in 2005 initiated a registration based census in which there was no comprehensive enumeration, but rather the analysis of other records of government combined with sample surveys. Denmark and Finland have population registers and these are combined with data from other administrative records. In 2004, France introduced a system in which data are gathered in successive annual collections rotating through municipalities every 5 years; these have replaced the comprehensive census. The American Community Survey (ACS) is administered annually to a sample of areas throughout the United States, with no household being sampled more than once in 5 years. Although the 2010 census will still take place, the data collected in the ACS will replace the long questionnaire used in previous censuses. Such approaches are likely to be adopted in other countries where registration systems are comprehensive, but may not be possible in many other countries with less comprehensive systems and where certain privacy regulations are in place.

Objectives and Design of a Modern Census

In many countries, there is legislation which regulates the implementation and analysis of the national census. A primary legislative function of many censuses is to allow equitable construction of electoral boundaries, and these are often redrawn after each census. Most censuses also have objectives related to planning for the provision of services such as education and health, the development of a skilled labor force, and a range of other economic and social purposes.

In order to fulfill these objectives, most modern censuses are designed to be:

• universal: the total population should be enumerated;
• compulsory: all residents are obliged to take part in census enumeration;
• comprehensive: ideally a large number of variables should be collected, although in reality some censuses collect samples for some variables (e.g., US census in 2000 obtained full enumeration for six basic variables but a 15% sample for a larger number of variables); and
• simultaneous: the census should take place within a short period of time for the whole population to allow comparability; there is often a precise census moment (e.g., midnight on a particular date).

The periodicity (frequency) of censuses varies considerably between countries. The US and UK undertake censuses every 10 years (in years ending in 0 and 1, respectively), and these are an artifact of the timing of the first censuses two centuries earlier. The French census has been carried out every 7 to 9 years, sometimes, but not always, approximating the presidential electoral cycle. Censuses in many ex British territories (Canada, Australia, and New Zealand) have followed a 5 year cycle, although the electoral cycle in those countries is shorter than this. In the attempt to achieve universal and synchronous enumeration, the timing of a census is important. It should avoid periods of extreme weather (although Finland uses New Year's Eve), significant cultural and religious events, and periods of high levels of resident mobility.

Enumeration coverage (who should be enumerated) is also an issue. A de facto enumeration (e.g., UK) records people according to the place they are at census moment, while a de jure enumeration (e.g., US) records the usual residence of a person. The problem with the de facto approach is that it may incorporate many people who are not usually residents (e.g., tourists) and so may be less relevant to the planning and other purposes of the census. The problem with the de jure approach is that while it may reassign enumerated residents to their usual place of residence, this is not possible for usual residents who are outside of the country at the time of enumeration (unless they are intercepted at the border on their return). In some countries, both de facto and de jure data are collected and analyzed.

To encourage international comparability, the United Nations has proposed a standardized range of variables that should be collected in a census, including age, sex, birthplace, citizenship, language, education, economic status, family structure, fertility, and a number of others. In practice, however, a great range of different information is collected by different countries. Although most censuses aim to be as comprehensive as possible, the number and nature of variables collected depends on financial and human resources available for enumeration and analysis, as well as political, social, and cultural considerations. Also, in countries with large sectors of the population with limited levels of education, it is necessary to create questions which are readily comprehensible to both enumerated and enumerator. Political considerations include the desirability of convincing the population that questions are not designed to allow agencies of government to use census information for purposes of cross checking on aspects such as income, welfare benefit payment, and military service. Social and cultural considerations include sensitivities about questions perceived as private such as questions relating to personal behavior or characteristics such as fertility, smoking, disability, or religion. In some censuses, questions such as these are optional.

Role of Government and Politics of Participation

Even in those countries which have implemented neoliberal reforms in which some functions of government have been privatized, the central statistical agencies which implement censuses have remained under government control (although often with a legislated degree of autonomy). The role of government is critical to the implementation of a compulsory census, but this role also creates problems in relation to participation.

In the past, it was mainly small religious groups which objected to being enumerated in line with their aversion to participation in most aspects of government. However, the politics of participation has become more complex in recent times with objections to the census process also originating from political movements such as libertarianism and anarchism. Also, as marketers, market researchers, credit rating agencies, and spammers of all varieties have become increasingly intrusive using new technologies, census enumeration may be classed by some along with these other intrusions. On a general level, many governments have responded to these intrusions by strengthening privacy legislation, and more specifically by increasing the provisions for confidentiality within the census collection and dissemination processes. Most recent censuses have been preceded by publicity campaigns emphasizing the public good aspects of a census and the strict provisions for confidentiality. Nevertheless, in even the most successful of censuses there are individuals within the population who will be missed (and occasionally counted twice), and other errors related to individual and household characteristics will enter into the census process. Thus, it is necessary to consider the process of census enumeration and analysis in more detail.