12.7 Household master sample, civil registration and vital statistics and population registers#
Having a reliable survey frame is essential to any survey. Even though NSOs benefit from using census information to construct household survey frames, updating the survey frame is often costly and complex. Depending on available resources and administrative environments, three general approaches to household survey frames construction can be identified:
Use of household master sample – commonly used by NSOs that has a sufficient number of independent surveys or periodic rounds of the same survey to sustain their use;
Use of population census – used by NSOs that have sampling experts but do not have the resources or administrative sources for regular and complete updating of the survey frame;
Use of population registers – used by NSOs that regularly update the complete survey frame via defined procedures from administrative sources.
Apart from the approaches mentioned above, this section will discuss the civil registration and vital statistics system, both as a source of official statistics and its usage to update the sample frame.
12.7.1 Household master sample#
A master sample frame is used for sample selection in a household survey, covers the entire target population and is used for multiple surveys or multiple rounds of the same survey. NSOs usually create a master sample frame after they finish the census processing by eliminating the information that is not needed for sampling purposes and making it available to sampling experts. Depending on security and data access policies, different approaches to the technical flow of individual and contact information are followed, but the general recommendation is that access to information should be closely guarded and limited.
The main advantage of a master sample design is that it allows one or more stages to be combined or shared among different household surveys. This approach reduces the costs of developing and maintaining sampling frames as more household surveys share the same master sample frame and frame design. It also simplifies the technical process of drawing individual samples and facilitates operational linkages between different surveys. Being a master sample frame, the sampling frame does not vary either from one survey to the other or from one round to another of the same survey. Instead – and this is its distinctive characteristic – the master sample frame is designed and constructed to be a stable, established basis for selecting the sub-samples needed for the particular surveys or rounds of the same survey over an extended period of time.
Many countries face budgetary constraints and rely on technical assistance from international development agencies. To promote the sustainability of statistical data collection activities, different strategies have been proposed to economize on the technical and financial resources needed for conducting household surveys. One of these strategies is the development of a master sample design.
Using a master sample design to select samples for multiple household surveys provides an opportunity to minimize the resources needed to collect household survey data regularly.
In particular, using the same sampling design and frame to select samples either for multiple surveys of different content or for different rounds of the same survey can produce significant cost-savings compared to developing an independent sampling design each time a household survey is carried out. Even though this strategy does not reduce the overall burden for respondents, respondents who live in a geographical area selected during the first stage may feel disproportionate response burden, particularly as surveys will not be collected in other areas.
A master sample is a sample drawn from a frame covering a target population for use on a number of future occasions, so as to avoid ad hoc sampling on each occasion. Sometimes the master sample is large, and subsequent inquiries are based on a sub-sample taken from it. The main advantage of using a master sample is that it lowers the cost of sampling as it provides a well-defined set of statistical units, which can be suitable, for instance, for panel surveys.
On the other hand, a master sample is a preselected set of units, which might cause bias when used as the source for a sample survey. Countries with good registers are increasingly moving to samples of individuals and use telephone or web interviewing in combination. The benefits of a master sample are not obvious in that case if the sources are regularly updated, and the sampling expertise is available in-house.
The terms ‘master sample frame’ and ‘master sample’ can be confused, especially if a multi-stage sampling design is involved. The defining characteristic of a master sample frame is that it serves as a sampling frame for several samples; for different surveys or different survey rounds. The defining characteristic of a master sample is that it is a sample, and different sub-samples (for different surveys or different survey rounds) are drawn from it. It is important to keep in mind that not every sample drawn from a master sample frame is a master sample.
In general, a master sample consists of an initial selection of primary sampling units (PSUs) that remain fixed for each subsample. Note that the latter stages are usually variable. For example, in the final stage of selection, the particular households chosen for interviews are usually different for independent surveys, while they may be the same or partially overlapping in repetitive surveys.
National household survey capability programme - Sampling Frames and Sample Designs for Integrated Household Survey Programmes (🔗);
Hans Pettersson - Design of master sampling frames and master samples for household surveys in developing countries (🔗).
Sources, creation and maintenance procedures
As noted in Chapter 9.2.4 — Survey design, in Chapter 12.3 — Statistical Business Register and in Chapter 12.4 — Frames for informal sector surveys, a business survey frame is derived from an SBR or in the absence of that, an economic census and the same applies to master sample frames for household surveys. A consistent survey frame is recommended for surveys with the same target population or a subset of the target population.
From a master sample frame, it is possible to select the samples for different surveys entirely independently. However, in many cases, substantial benefits result from selecting one large sample, a master sample, and then selecting sub-samples of this master sample to service different (but related) surveys. Several NSOs have decided to develop a master sample to serve the needs of their household surveys.
Over the time period during which a master sample frame is in use, changes will occur that affect both the number and the definition of the frame units. A master sampling frame is basically a list of area units that cover the whole country. For each unit there may be information on urban/rural classification, identification of higher-level units (administrative subdivisions – i.e., district and province to which the unit belongs), population counts and other characteristics. For each area unit, there must also be information on the boundaries of the unit. The amount of work required to maintain a master sampling frame depends primarily on the stability of these frame units. There are two kinds of changes that may occur in the frame units: changes in frame unit boundaries and changes in frame unit characteristics. Changes affecting the boundaries of frame units must be recorded in the master sample frame, and a system for collecting information about administrative changes needs to be established to keep track of these changes. Changes in frame unit characteristics include simple changes such as name changes and more substantial changes like changes in the measure of size (population or number of households/dwelling units) or changes in urban/rural classification. These changes do not necessarily have to be reflected in the master sample frame. However, as has been said above, outdated information on measures of size results in a loss of efficiency in the samples selected from the frame. Updating measures of size for the whole frame would be very costly and generally not cost-efficient; but for especially fast-growing peri-urban areas, it is a good idea to update the measures of size regularly.
There are examples where a master sample frame is periodically or continuously partially updated from CRVS systems (if the master sample contains individual information) and other administrative sources (such as local government registers, lists of voters, lists of immigrants and emigrants…) to reflect deaths, births and migrations. Also, there are examples where the update of the master sample is performed through updating address information from administrative information or via area canvassing. This is rather expensive and therefore rarely used. Examples where canvassing are focused on newly constructed buildings identified from geospatial (or other) information. Even though that periodic updates are beneficial, a full update of the master sample frame should be performed as soon as census microdata becomes available.
Generation of survey frames and samples
As noted earlier, a master sample is a large sample from which different sub-samples are drawn for different surveys. The subsampling can be carried out in many different ways. Subsampling on the primary level (of primary sampling units) would give a unique subsample of the master sample primary sampling units for each survey, as each survey would have a different sample of enumeration areas. Subsampling on the secondary level would give a subsample of housing units from each master sample primary sampling units, as each survey would have the same sample of enumeration areas but different samples of housing units within the enumeration areas.
The subsampling could be carried out independently, or some kind of controlled selection process could be employed to ensure that the overlap between samples is at a desired level. Another way of selecting samples from the master sample is to select independent replicates from the sample. One or several of the replicates could be selected as a subsample for each survey. Such a set-up would require that the master sample be built up from the start from a set of fully independent replicates.
ST/ESA/STAT/SER.F/98 - Designing Household Survey Samples: Practical Guidelines (🔗);
Maligalig, D. S., & Martinez, A. Jr (2013). Developing a Master Sample Design for Households Surveys in Developing Countries: A Case Study in Bangladesh (🔗);
ESA/STAT/AC.93/3 - Sampling frames and master samples - Anthony G. Turner (🔗);
UNSD - Household Sample Surveys in Developing and Transition Countries (🔗);
ESSnet KOMUSO – Quality Guidelines for Frames in Social Statistics (🔗).
12.7.2 Civil registration and vital statistics systems#
Ideally, Civil Registration and Vital Statistics (CRVS) is the continuous, permanent, compulsory and universal recording of the occurrence and characteristics of vital events of the population. The actors in a CRVS system typically include the civil registration authorities, Ministry of Interior or Home Affairs, Ministry of Justice, Ministry of Health and the national statistical office.
CRVS provides an important administrative function as it records documentary evidence of life events and family relations. Functioning CRVS systems record every birth, adoption, marriage, divorce and death, alongside additional information – such as the cause of death.
Recording every vital event is also an opportunity to produce the most accurate, complete and timely statistics on the population’s health and demographics and maintain a national civil registry or a population database. While producing vital statistics through sample surveys serves a valuable function of highlighting and drawing attention to the possible scale of the situation, these can be costly, uncertain and may not cover the most vulnerable segments of the population, nor can they be localized. They are not a long-term substitute for vital statistics derived from a universal CRVS system. Even though CRVS systems are in place in most of the world, a substantial part of the world population did not get registered at birth in a civil registration system. UN Statistics Division has initiated the formation of the Global Civil Registration and Vital Statistics Group (🔗), a group of international and regional organizations to strengthen and support the establishment and modernisation of national CRVS systems. Similarly, Statistical division of United Nations ESCAP has launched a CRVS promotion campaign Get into the picture (🔗) that is promoting the establishment of CRVS systems in Asia and the Pacific.
At the end of May of 2019, the United Nations Legal Identity Agenda was launched in Rome on the Conference in preparation for High-Level Political Forum: Peaceful, Just and Inclusive Societies – SDG 16 implementation and the path towards leaving no one behind. The United Nations Legal Identity Agenda refers to the holistic approach to civil registration of all vital events, production of vital statistics, the establishment and maintenance of population registers and identity management apparatus from birth to death, and there should be full interoperability[1] between these functions in a simultaneous manner, according to international standards and recommendations and in compliance with human rights of all people concerned, including the right to privacy. All Member States should adopt and implement this agenda as a systematic and perpetual mechanism for ensuring legal identity for all. The United Nations Statistical Commission adopted the UN LIA at its 51st Session in March 2020.
Apart from their use as a source of statistics on demographic events, CRVS systems (in combination with census and other sources) can be used to produce population estimates in inter-census periods and also population projections. Further, individual information from CRVS systems can be used to update the master sample frame information. Updating the master sample frame (as well as master sample) from CRVS systems is recommended, particularly for removing the deceased persons from the survey frame as this significantly reduces inconveniences for interviewers and consequently for NSOs.
Principles and Recommendations for a Vital Statistics System, United Nations, 2015 (🔗);
Handbook on Civil Registration and Vital Statistics Systems: Management, Operation and Maintenance Revision 1, United Nations, 2018 (🔗);
Guidelines on the Legislative Framework for Civil Registration, Vital Statistics and Identity Management, United Nations, 2019 Population registers (🔗).
Having a fully reliable CRVS system and using its individual information to update the sample frame is not enough to get the complete sample frame in line with the total population. Even though information on life events is present, CRVS generally does not track information about the residence of individuals. Even if such information is present in CRVS or in a different administrative database, it is often not regularly updated or does not contain links between a person and a dwelling, thus making it difficult to update the individual information on households from the sample frame.
In some (mainly northern European countries, also Slovenia and Austria) individuals are obliged by law to register themselves administratively to a specific dwelling, and that information along with items (such as educational attainment, and current economic activity) is stored in an administrative population register. An administrative population register is usually, connected with dwellings and buildings register (described in Chapter 12.6 — Household address register) on individual level, as each unique person (with unique personal ID) has to be connected with a unique dwelling (with unique dwelling ID).
Having a reliable administrative population register greatly reduces the total costs of official statistics as it eliminates the need for field collection of information that is available from reliable administrative sources.
Countries with administrative population registers do not perform field collection of Census data but connect different administrative databases and may conduct a survey to collect information that enables imputation of missing data. Even though this approach is cost-effective and enables the annual production of otherwise available indicators every five or ten years, administrative data holders are generally concerned about the response burden, which generally leads to a limited number of available variables being collected. Further, the transition to administrative census opens the problem of comparability between previous censuses.
In the context of the United Nations Legal Identity Agenda (described above), the role of population registers extends well beyond both governmental and statistical functions and becomes a mechanism for ensuring the legal identity for all from birth to death by issuing legal credentials based on civil registration throughout the lifetime of an individual. The population register – accurate and updated regularly – represents a reliable source of information on population migration. In the context of generating small area census statistics, the establishment, maintenance and operations of a universal population register – combined with information synthesized from other functional registers – represents a significant and efficient alternative to costly canvassing of the whole country.
Consequently, it is of particular importance for NSO’s to be closely involved in the legislative process, establishment, operations and protocols to access the information from population registers for statistical purposes from the onset of such a process in countries. While establishing the statistical exploitation of population registers at the later stage would not be impossible, it certainly would require more adjustments than if those relationships are elaborated at the very beginning.