1 . Full field enumeration (traditional census)

The full field enumeration or traditional approach comprises a complex operation. of actively collecting information from individuals and households on a range of topics at a specified time, accompanied by the compilation, evaluation, analysis and dissemination of demographic, economic and social data pertaining to a country or a well-delimited part of the country. Members of the public respond to a census questionnaire, or interviewers are deployed to collect information from respondents. For interviewer-based censuses, enumerators assigned to different enumeration areas cover all households and persons in the enumeration area during a specified and usually short period of time in order to meet the requirements of universality and simultaneity. Either a single long form is universally canvassed, or a combination of short and long forms used. In the latter case, the short form contains only questions intended for universal coverage, while the long form is used to collect information from only a sample of households and population. This form usually contains detailed questions on a particular topic in addition to covering complex topics such as fertility. Both forms are utilized during the same time frame of the census. While the long form estimates are not based on full coverage, they are regarded as census output. Overburdening the census form is likely to adversely affect response rates and the quality of data.

A variant of this approach consists in modelling the full field enumeration with a continuous cumulative survey in order to generate yearly (or at other intervals) estimates of detailed characteristics of population and housing. The primary advantage of this approach is to provide more frequent and relevant data on population and housing than would be available when a census is conducted only once a decade. However, such a program might be costly and technically difficult, as it requires a multi-year-round of comprehensive planning, development, and testing.

As various methods can be used for collecting the data, including a mailed or dropped off questionnaire, the telephone, the Internet, personal visit follow-up, or a combination of such methods, countries employing the traditional design may utilize very different collection approaches in doing so.

The traditional census has merit in providing a snapshot of the entire population at a specified period and data for small geographic domains. In that sense, the traditional census is perhaps unique in nature. This approach is particularly suitable for countries requiring population numbers by various social and economic characteristics simultaneously for all geographic levels to meet the needs of planning and the allocation of funds. The delimitation of electoral boundaries requires simultaneity, and for that reason also the traditional approach may be more appropriate. But at the same time, traditional censuses have been singled out as the most elaborate, complex and costly data collection activity that national statistical authorities undertake. In addition to costs, this complex task requires full awareness and agreement of the public to participate in it.

1.1 - Necessary conditions

It is essential to have national legislation for conducting the population and housing census to ensure confidentiality, transparency and the cooperation of the population. A permanent central census organization, which may or may not be part of the statistical office, needs to exist in the country, which can be expanded during the time of the census. Since a traditional census requires substantial resources, sufficient funding for a field operation covering the entire country and subsequent data processing needs to be ensured. Other conditions necessary for this approach are the support of and general acceptance by the public to participate in the enumeration, and trust towards the statistical office or census agency.

2.1 - Advantages and disadvantages

The two biggest advantages of a traditional census are comprehensiveness of coverage and simultaneity. Another major advantage is the flexibility in deciding the topics to be covered and design of the questionnaire. There is lesser need for complex data adjustment since processing of raw data provides all inputs. The census frame becomes the base for all subsequent sampling frames. Finally, the focused and time-bound nature of the field operation implies that the data collection is finished in a short period and does not require long term or constant monitoring.

One of the biggest disadvantages of a traditional census is its cost and administrative complexity. Another disadvantage is that it has a very long processing time. Also, since it can be conducted only after 5- or 10-year intervals, data tend to become outdated. The burden on respondents can be considered a disadvantage, especially in countries where participation in the enumeration is declining. Finally, many countries experience increasing difficulties in enumerating specific population groups, such as persons with high mobility or with multiple residences, or who are difficult to reach for other reasons.

Some considerations for census taking and content

Very precise planning is required for every stage of census taking in the traditional approach, due to the sheer volume of work and overlapping time frames. The recruitment and training of a large number of census takers adds to cost and complexity. Involvement of administrative machinery at the central, provincial and local levels is essential for successful field operations.

Since the data obtained in the traditional method are respondent- and enumerator based, there is scope for error in canvassing the questions and in the quality of response. This, however, can be minimized through proper design of the questionnaire, effective training and wide publicity. Data can be provided for every administrative level subject to privacy and confidentiality considerations, which may not always be possible with other methods if some parts of the data collection are based on sample surveys.

2 . Register-based census

The concept of producing census-like results based on registers developed in the 2000 round of censuses, although it has been debated and tested to various degrees since the 1970s, and several countries succeeded in using this approach to generate census data in the 1990 round of censuses. The philosophy underlying this concept is to take advantage of existing administrative sources, namely different kinds of registers on individuals, households and dwellings. These registers are linked at the individual record level with information held on business, tax, education, employment and other relevant registers. While it is theoretically possible to link records on the basis of the name and other unique details of the individuals, the existence of a unique identification number for each individual, household and dwelling allows a much more effective and reliable linkage of records from different registers.

Administrative registers are produced on the basis of administrative processes to collect information on units (persons and housing units) and variables that are defined by administrative rules and demands in a country. Although the content and process of registers would differ from one country to another, the types of the registers are usually very similar. The following provides definitions of the main concepts used in the system of administrative registers.

(a) A register is defined as systematic collection of unit-level data organized in such a way that updating is possible. Updating is the processing of identifiable information with the purpose of establishing, updating, correcting or extending the register.
(b) Administrative registers are registers primarily used in an administrative information system. This means that the registers are used in the production of goods and services in public or private institutions or companies, or that the information is a result of such production. Administrative registers used for statistical purposes are normally operated by the State or jointly by local authorities, but registers operated by private organizations are also used.
(c) Administrative base registers are kept as a basic resource for public administration. The function is to keep stock of the population and to maintain identification information. Statistical base registers are based on the corresponding administrative registers. Their principal tasks are to define important populations and contain links to other base registers.
(d) Specialized registers are registers that, unlike base registers, serve one specific purpose or a clearly defined group of purposes. Specialized registers often receive information on the population and some basic data from a base register, but supply other data themselves (such as the register of vehicles).
(e) Statistical registers are created by processing data from administrative registers for statistical purposes. A statistical register could be based on one or several administrative registers. Statistical registers are also referred to as secondary registers.

The process involves collecting information on the characteristics of individuals, which has been provided to an administrative register for non-statistical purposes. To be effective, access to administrative data for statistical purposes must be given by law or by agreement, providing the capability to (a) transfer the data as individual records to the statistical database; or (b) temporarily link the registers to form a proxy register for statistical purposes.

Administrative registers are maintained primarily for administrative purposes. Units and variables of administrative data are described according to administrative rules and demands. Before a register is used for census purposes, the suitability of its data in terms of definitions, concepts, content, reference date, accuracy and other criteria should be statistically tested by comparing them with previous census and survey results, and conducting quality and compatibility surveys. A pilot census may be used for this purpose. Some backbone registers and administrative sources are:

(a) Population register (base—usually covers births, deaths, marriage and migration);
(b) Buildings and dwelling or address register (base);
(c) Business register (base);
(d) Taxation register;
(e) Employment register;
(f ) Pension register;
(g) Social welfare register;
(h) Jobseeker register;
(i) Student register.

All persons within the defined territory who meet the register’s rules are enumerated. In concept, the enumeration is taken from a population register in which the fields for different census attributes are populated from subsidiary registers relating to specific topics. Information is extracted from the register as it reflects the situation of individuals at the predefined census reference date. The timing of the census extraction may require careful thought where register update cycles vary. Registration delays and administrative delays in updating between regional and national databases can otherwise have a serious impact on the quality of the output.

1. Register source with existing sample survey

A special case of register-based census is when population and/or integrated administrative registers are combined with existing sample survey(s). Different data sources are integrated as part of a “virtual census” process. The data for the census exercise are derived from many types of registers and surveys, covering different population groups throughout the country and its subregions. Compared to conventional census methods, this process is lower in cost and staff requirements, and uses more frequently updated data sources. More significantly, by combining data that are already available from other sources, the virtual census makes no additional respondent burden, increasing its public acceptability.

A weakness of the use of sample data within the census exercise is that for the variables derived from the sample surveys the sample size may not be sufficient to offer the geographically detailed outputs that are an essential feature of the census. Importantly though, the use of data from existing regular surveys will often mean that time series of data are available. By combining samples for several survey exercises, it may be possible to produce reliable estimates for small geographic areas.

However, the use of existing surveys does present a number of problems compared to the use of ad hoc surveys. The timing, content, statistical definitions and sampling approaches used in an existing survey may not be appropriate to allow the data to be readily combined with data from the administrative source(s). For example, many major household surveys are not designed to cover persons living in institutional households (such as student accommodation, hospitals, prisons and military establishments), meaning that an additional source of information is needed for these persons.

2. Necessary conditions

Among the essential preconditions to conduct a register-based census is that the country should have:
(a) A national legislation providing for the creation of a population register and permission to use the data contained in it for statistical purposes;
(b) An established central population register;
(c) High-quality data in the population register;
(d) Comprehensive geographic coverage in the register;
(e) An effective system of continuous updating of the population register.

With regard to other administrative registers used, the following are essential:
(a) Access to data in the various registers should be allowed through legislation;
(b) The concepts and definitions used in the various registers should be harmonized;
(c) A universal personal identification (unique identity) system should be in place to facilitate proper linking of data;
(d) Quality and consistency checks should be conducted to verify the suitability of the data contained in various registers.

3. Advantages and disadvantages

The primary advantages of a register-based approach are reduced costs and greater frequency of data. However, establishing and maintaining administrative registers involve higher costs than the census alone may justify. The need for the register will largely be based on its contribution to more reliable and efficient administration. The use for statistics may be valuable but is likely to be a secondary consideration.

Certain potential drawbacks with the use of administrative data sources also need to be considered. One limitation is that the scope of statistical topics, key definitions and, indeed, the population base of the exercise depend on the information that can be compiled from the available registers. These, in turn, will be based on the underlying administrative purpose and procedures of the registers. In addition, it is common for national legislation to restrict or prevent the use of administrative registers for other purposes, including the production of statistics. This may impose restrictions with respect to the characteristics that are available for description and may also undermine international comparability. When a registered data item is changed, new or updated information is not always registered immediately. In certain cases, new or updated information may not be registered at all. Where this occurs, the register information does not accurately reflect real circumstances.

4. Some considerations for census taking and content

Administrative registers can, depending on content and quality, be used in all phases of census-taking. In principle, where greater amounts of information can be obtained from administrative sources, the production of census-type statistics will be faster, cheaper and more complete. The most complete use of registers will be where all core, and in some countries non-core, census topics can be based on register information. It is possible to improve the quality of data collected through the administrative register by conducting a sample survey or surveys. The sample survey(s) may either use the register as a sampling frame, or else be completely independent of the register.

The use of register data may reduce the flexibility of the census exercise in terms of the variables that are available and their definitions. It may be difficult to change the variables as these are defined in line with administrative priorities. A significant potential risk for the success of the census exercise is that the administrative source will often be outside the control of the statistical authority. The influence of the statistical authority over the administrative source can be very limited. The content and availability of the administrative source may change at relatively short notice and without reference to statistical needs. For example, a change in taxation legislation may mean that a key administrative register may no longer collect information needed for the census. This risk can be minimized by establishing close and regular communication between the statistical authority and the owners of the administrative sources.

In practice, only persons legally present in the various registers would get covered through this approach. For example, unregistered births, deaths or marriages, illegal immigrants, homeless persons, nomadic or floating populations, persons involved in illegal activities, etc., are not likely to be recorded in any such administrative register. On the other hand, registers may include persons who are not living (any longer) in the country, for example persons who emigrated but were not cancelled from the registers.

Summing up, subject to the caveats mentioned in the paragraphs above, the feature of individual enumeration is satisfied in this approach as separate information is collected regarding the characteristics of everyone. As regards “universality within a defined territory”, this criterion is satisfied as the enumeration is taken from a population register in which the fields for attributes are populated from subsidiary registers relating to specific topics. Regarding “simultaneity”, the timing of the census extraction may require careful thought where register update cycles vary. With respect to “periodicity”, this approach allows extraction at desired frequencies, including “at least once in 10 years”, noting again the need to manage the updating cycles for the registers.

Finally, in most cases, the requirement in terms of producing small-area statistics is largely met, as the information in the registers allows for such aggregates to be generated.

3. Combined methodologies.

In recent years, it has been observed in several countries that:
(a) The quality of the administrative registers is relatively good (at least for certain key census variables).
(b) Information for some census topics is not available in the administrative registers or the quality is not sufficiently high.
(c) The population generally, and certain population groups (in particular people difficult to enumerate), are becoming more sensitive to the handling of personal information, and possibly more reluctant to cooperate with the statistical office or more difficult to enumerate due to their high mobility or other reasons.

In these cases, a combined census that uses register(s) and questionnaire(s) could be an option. Essentially, the combined methodology makes use of registers relevant to a census, complemented by surveys or complete enumeration. The use of survey and enumeration data is intended to:
(a) Improve the accuracy of the population counts;
(b) Provide information for census variables that cannot be reliably based on administrative data;
(c) Check, update and improve the quality of census data derived from administrative sources;
(d) Add additional variables to the census;
(e) Be a linking frame in order to bring together different sources.

Information on individuals, households and dwellings is collected by combining data from registers with data collected from one or more surveys. Data collection may be based on full field enumeration, an ad hoc sample and rolling survey methods. Data from registers are employed not only as a sampling frame or to support field operations, but also directly as the data source for some census information. In a case where registers are used along with total enumeration, data from registers may be prefilled in questionnaires, and respondents may be asked to check, update and confirm their details. Other questions relating to fields not available in the registers may also be canvassed during this exercise. In the case of new individuals, households or dwellings that do not feature in the register, all fields of information that are required for the register and the census are to be canvassed afresh. When registers are used along with sample surveys (ad hoc sample or rolling surveys), some census tables may be produced entirely from the information available on the register(s), while for other census tables, information from the survey(s), duly weighted to the population totals, could be used. The surveys would also serve to evaluate the accuracy of the register counts.

1. Necessary conditions

The option of a register-based census with sample surveys can be adopted only if all necessary census information is available from the various administrative or survey sources, and it is possible to link the information from the different sources at the record level. During the process of integrating individual records, care should be taken to check the accuracy of the data and remove inconsistencies prior to the production of statistical outputs.

The data sources would include verified and accurate personal information (name, ID number, date of birth, sex, marital status, family structure, etc.) and a dwelling register. In an ideal situation, a “base” register can be envisaged, to include unified identity codes for both people and address components to link more efficiently the related register and survey data. The link between persons and their dwellings is equally important, giving the household unit. Other administrative data sources include tax files, social security files, public records of unemployed and registers of educational qualifications. It is preferable to have a centralized base register. If this is not available, regional registers will need to be consolidated.

2. Advantages and disadvantages

The register-based census combined with full enumeration or surveys offers several advantages:
(a) It can be much cheaper than a traditional census with a full enumeration collecting all census items from the whole population;
(b) It will reduce the burden of enumerators and respondents;
(c) It will reduce non-response in case information is obtained from registers;
(d) It should be possible to correct the survey data for differing levels of non-response in different population groups.

Micro-integrated data might be expected to provide very reliable results, because they are based on a maximum amount of information. The coverage of subpopulations may be more reliable because when data are missing in one source, another source may be used. Another advantage of micro-integration is that there will be less reason for confusion among statistics users. For example, there will be one figure on each socioeconomic phenomenon, instead of several different figures depending on which sources have been used.

A disadvantage is that it involves more work to produce the tables from the sample survey microdata, as weighting problems may arise. As the combined census may lack the high public profile and publicity of a traditional census, there may be less interest in and use of the census results, as there is no longer a single census event to attract public attention. Other potential disadvantages may be a lack of transparency (no one external to the process may be able to reproduce the information) and data quality.

3. Some considerations for census taking and content

Data validation, processing and dissemination may be more complicated, as this approach involves both total counts based on the register and sample data from surveys. In addition, as some variables are based only on sample data, it may be impossible to meet the level of statistical and geographic detail required in some tables. On the other hand, the possibilities of reducing cost and response burden provide a very strong reason to adopt this approach.

Some of the required variables will need to be constructed from different sources. The census results obtained may differ to some extent from those that would be obtained from a full enumeration covering all census topics. This may have a negative impact on the comparability of results between countries and over time. An advantage of registers is that, in effect, they offer complete coverage subject to the quality of the data contained. It is preferable that statistical authorities make full use of the register data that are available.

For the combined census method, several different methods can be used to collect information, including paper, Internet, handheld devices and telephone interviews. Electronic devices have important advantages that influence the quality of the information obtained: validation controls can be included in the different questions, time to answer the questions is reduced and the analysis and dissemination of information is faster. Where data are collected via different routes (such as Internet collection in parallel with face-to-face interviews), controls are needed to avoid duplication of information.

4. Examples of combined methodologies

4.1. Non-continuous approach
(a) Base register and/or integrated administrative sources with full field enumeration.
One approach can be to combine the full enumeration with a base register or registers. The questionnaire used in the total enumeration then contains fewer questions compared to a traditional census questionnaire, but still covers the whole population of individuals, households, and dwellings. Over time, countries may decide to adopt this model, increasing in successive censuses the use of integrated administrative registers and reducing the number of questions in the questionnaire. The registers can be used to prefill such information as name, address, family composition, education, occupation and dwelling characteristics on the questionnaire. These prefills can then be used to ask the respondents if the information is still valid or needs to be changed. That can greatly reduce the work involved in coding of the census questionnaire.

Advantages and disadvantages
The main advantage with using this model is that it will reduce the response burden for respondents and reduce the cost of the census. The model also allows the preparation of small-area statistics as all variables are collected as total counts. However, this model will still involve a large data-collection exercise with the use of enumerators. Mail out/mail back data collection may sometimes be used, but a significant proportion of respondents may require enumerator follow-up and assistance in completing the questionnaire.

(b) Base register and/or integrated administrative sources with ad hoc sample survey(s)
Another model involves the use of an ad hoc sample survey instead of a full enumeration. The backbone register could then be used as a sample frame and also to prefill some information such as name and address on the questionnaire. The ad hoc sample survey questionnaire can be specifically designed to complete and statistically correct the data coming from registers, covering those variables not available from the register. The sample can be sized and stratified in such a way that data are available for small groups and geographic areas.

Advantages and disadvantages
This model requires far fewer enumerators than a full enumeration, so a more specific training operation with skilled and prepared professional interviewers can be carried out. The follow-up of the operation is also simpler. There is no need to obtain information from each member of the population, giving a clear reduction in response burden. Non-response can be corrected in the sample using statistical techniques to ensure information is still representative of the population.

By using a sample and a much smaller number of enumerators, a significant reduction in the cost of the census is possible, in particular when compared with the traditional approach.

The extent to which this model can produce detailed statistics (in particular for municipalities or smaller geographic areas) will depend on the size of the sample survey. A larger sample should allow more detailed statistics to be produced but this will correspondingly increase the financial cost of the exercise. Even with a large sample, the availability of detailed information is likely to be lower than if a comprehensive approach were taken—such as with a traditional census or fully register-based exercise.

4.2. Continuous approach

(a) Rolling census
In a rolling census, information is collected on individuals, households and dwellings by a continuous cumulative survey covering the whole country over a period of time (generally years), rather than a particular day or short period of enumeration. The two main parameters of a rolling census are the length of the period of enumeration and the sampling rate (which depends on the geographic levels required for dissemination purposes). For example, it is possible to build a sample framework in order to produce national results with one annual survey, regional results by cumulating three annual surveys, and small-area results by cumulating data over five years. Annual surveys may be conducted over the full course of the year or in a particular month or other shorter time frame.

Necessary conditions
The necessary conditions partly depend on the complexity of the sample framework. If the sampling units are addresses, a master address file is to be built first. But if the sampling units are larger, for example municipalities, it is only necessary to have enough information to spread the municipalities over the different years. It will be necessary to explain to statistics users the impact of the rolling sample on the use and interpretation of data, as many users are more used to snapshot data rather than period data.

Advantages and disadvantages
The main advantage of the rolling census approach is the higher frequency for updating data: a traditional census provides benchmarks every five or, more commonly, ten years. In contrast, the rolling census provides annual updates. Another advantage is the reduction in the burden on the public. The high peak costs and labor requirements of a traditional census are instead spread over a longer period. Furthermore, it is possible to improve the census process over time, and to test methodological refinements and new technologies as they emerge.

The disadvantage is that the rolling census approach no longer provides a snapshot of the whole population, complicating comparisons between areas due to different enumeration times. In addition, as the rolling census covers the whole country over a period of time, some respondents will move. Thus some people may be surveyed several times and others may not be surveyed at all. As a result, universality might not be ensured unless careful methodological adjustments are made.