New data sources can make censuses more efficient – UNECE issues guidelines to help maintain high quality

New Guidelines issued today by UNECE

will support countries in maintaining high quality as they move towards using modern data sources in their censuses.

Knowing the number of people in a population, their distribution across the country, and their basic characteristics such as age, sex and socioeconomic conditions is crucial for the good functioning of governments, as well as for businesses, researchers and individuals to make informed decisions.

The way that these core population statistics are produced is undergoing a worldwide transformation.

A population and housing census is conducted in most countries every ten years. But gone are the days when a majority of countries conducted their census by knocking on every door and interviewing every household. While some countries do this, a range of other techniques is now employed across the UNECE region: from mail-in or online forms, through combined approaches that integrate various sources with short questionnaires or sample surveys, to censuses based entirely on population registers, so that the members of the population don’t have to complete any questionnaire at all.

No matter which of these techniques is employed, the high costs and intense burden of conducting a census are increasingly being eased by use of administrative data sources.

Administrative sources are data that are not collected principally for statistical purposes-they can include tax records, electoral rolls, health system data, business registers, social security records and lists of addresses from commercial mail carriers, among a great many others. These sources can be used to support censuses in a whole variety of ways.

  • Countries that carry out a traditional census, where a paper or electronic questionnaire is sent to households, can use administrative sources to help them identify all the homes and individuals that should be enumerated; to efficiently deploy interviewers; or to select a representative sample to receive a long-form questionnaire. In the United States, for example, a range of administrative sources are used to predict whether an address is occupied, ensuring home visits are efficiently targeted.
  • Administrative sources can provide a benchmark against which to check the accuracy or coverage of data collected through questionnaires-and potentially to correct errors, in some countries. In New Zealand’s 2018 census, administrative sources were used to fill gaps where the field collection had resulted in undercoverage.
  • Countries embarking on a gradual shift away from asking householders a full array of questions might use administrative sources to add new information, including things that could be hard to ask people. For example, in some countries it might be sensitive to ask about infant mortality, or about the incomes of other household members. People might not wish to answer, may not know, or may give inaccurate responses. In the census of England and Wales, administrative data gathered for property valuations has been used to derive information on the number of rooms in homes, as respondents’ answers to questions about this were found to be unreliable.
  • Even further along the journey towards a streamlined, register-based approach to conducting a census, there are countries (such as Spain and New Zealand) which use these administrative sources to build statistical registers which are then used to produce census statistics.
  • Finally, a growing number of countries are following in the footsteps of the frontrunners – the Netherlands and the Nordic countries-which have for decades benefitted from complete registers of the population, including where they live, where they work or go to school, and other information needed to produce census statistics.

Each of these kinds of use comes with its own risks and challenges. Assessing and assuring the quality of the data sources used, and of the resulting statistics, is also essential to safeguard the public’s trust in the methods used. People may understandably have reservations about statistics being produced from these non-statistical sources, without ever being asked questions about themselves. National Statistical Offices (NSOs) are guided by data protection and confidentiality legislation, and have to be transparent about the methods they use to obtain and combine data, the safeguards they put in place, and the confidence we can have in the resulting figures.

Drawing on quality frameworks and best practices adopted by NSOs across the world, the new Guidelines lead census practitioners through the practical stages of assessing the quality of administrative sources; from working with an administrative data supplier to understand the strengths and limitations of a source, all the way to the receipt and analysis of the actual data.

The Guidelines cover key quality dimensions on which an assessment is made, using a variety of tools and indicators. The Guidelines are based on four Stages: Source, Data, Process, and Output, with the first two Stages being the principal focus of the Guidelines.

The Guidelines are the result of work undertaken by a Task Force of experts from 20 countries and organizations, reflecting the wide diversity of census methodologies and uses of administrative data in the UNECE region. The experiences of many of these countries are included throughout the Guidelines as illustrative examples and case studies, showing how the tools presented can be put into practice.

The Guidelines were endorsed by the 69th plenary session of the Conference of European Statisticians in 2021. They form part of a suite of UNECE guidance on censuses, including the 2018 Guidelines on the use of registers and administrative data for population and housing censuses and the decennial Conference of European Statisticians Recommendations for Censuses of Population and Housing, for which a project has just been launched to produce a revised version for the 2030 census round.

/Public Release. This material from the originating organization/author(s) may be of a point-in-time nature, edited for clarity, style and length. The views and opinions expressed are those of the author(s).View in full here.