Symposium 2001/26

11 July 2001


English only


Symposium on Global Review of 2000 Round of

Population and Housing Censuses: 

Mid-Decade Assessment and Future Prospects

Statistics Division

Department of Economic and Social Affairs

United Nations Secretariat

New York, 7-10 August 2001













Statement from Israel *

Eliahu Ben-Moshe **



A. On the definition of census success. 2

B. The “administrative” versus the “statistical” approach to censuses. 3


1.                  In this brief statement I focus on one aspect related to the main issue around which this Symposium has been organized: the definition of census success. I will claim that we should shift our approach to censuses from the “administrative” approach to the “statistical” approach. In this way we may be able to discover new ways to achieve census goals.

A. On the definition of census success

2.                  "When and under what circumstances do censuses succeed?" It is my understanding that the success of a population and housing census should be measured by the extent to which the census achieves its main and unique goal:

To provide data on the:

·        demographic and socioeconomic characteristics

·        at a specified point in time (and with well-defined periodicity)

·        of all the individuals and households living in a country (or a well-defined part of it)

·        that will allow constructing reliable estimates of these characteristics for small population subgroups (like populations in small areas).

3.                  In the enunciation of the census goal I stressed three of its traditionally accepted essential characteristics: universality, simultaneity and defined periodicity.

4.                  At the same time I intentionally excluded one characteristic traditionally considered essential to the census: individual enumeration. I did it because, in my understanding, individual enumeration should be considered a “means” and not a goal in itself. We do not produce individual data as an output; we just use these data to create different statistical aggregates. If we can find an alternative way to do that (i.e., by using small-area estimation techniques based on both surveys and administrative data), I can hardly see any reason to consider “individual enumeration” as an essential census characteristic.

5.                  I also stressed one additional characteristic: the ability of the census to permit constructing reliable estimates of the population and household characteristics for small population subgroups (or small areas). Small-area and small-population statistics have always been recognized to be one of the more peculiar and unique outputs of the census. Standard sampling surveys are by definition unable to provide reliable estimates for small populations, and individual administrative data sources are not able to provide integrated demographic and socio-economic data. In the best of the cases, they provide data related only to a specific subset of variables. Despite the wide agreement about the importance of small-area and small-population statistics as a census goal, they have not been in the past part of the “official” definition of a census. The reason is that since “individual enumeration” and “universality” were considered as essential census characteristics, small-population statistics could be computed in a straightforward way.

6.                  In practice, what we need is for the census to be able to provide reliable estimates of the population and household characteristics at a very detailed small-populations level that may be aggregated to any higher level, and this will be true as well for small population subgroups. Individual enumeration may not be needed for that. We should also remember that we already agreed in the past that some of the variables may be collected for a sample of the population and therefore we accepted that the estimates would be possible only from a given (small) population size, since the sampling error makes impossible the production of reliable estimates at smaller levels.

B. The “administrative” versus the “statistical” approach to censuses

7.                  For many (more or less historically justified) reasons, we tended in the past to define a census on the basis of its operations and not of its goal (“the total process of collecting, compiling, evaluating, etc.”). I claim in this brief statement that this was not correct and that the changing social and technological environments within which the census is conducted nowadays (and will be conducted in the future) are forcing us to review our approach.

8.                  It is clear that such an expensive operation like a census cannot be justified unless it meets its unique role. As mentioned before, its role is unique because it cannot be achieved using other alternative statistical operations like traditional sampling surveys or individual administrative sources. Once its goal is defined and agreed upon, any operation that will achieve that goal will be considered a census: a traditional census using any combination of enumerators, mail, telephone  and web-collection activities, an administrative census relying only on administrative data, a combination of administrative files data and sampling surveys or any other type of solution we may find in the future.

9.                  The need to define the census on the basis of its goal, and not on its operations, is of fundamental strategic importance for the future. The changes we are facing in the social milieu and public opinion that forced several countries to cancel their national censuses in the past and others to avoid the use of the word “census” when conducting census activities, the tremendous and fast-increasing financial burden of the census operations on the national budgets, and the new opportunities embedded in the new and fast technological and (less fast) methodological developments, are enough reasons for doing that.

10.              Changing our approach from the “administrative approach”, which stresses census operations that are more like administrative than statistical operations, and moving towards the “statistical approach”, which stresses the achievement of census goals, will bring us closer to finding more sophisticated and statistically oriented ways to achieve these goals.

11.              Important progress in this direction has already been made in the last census round recommendations (see the most recent recommendations for the 2000 censuses of population and housing in the ECE Region and also the United Nations principles and recommendations for population and housing censuses, revised in 1998). Also, several countries have already conducted or planned censuses that are innovations in the way to achieve census goals (see the US Bureau of the Census sampling plans for 2001 that unfortunately have not been implemented).

12.              There may be more than two main ways (traditional and administrative, or a combination of them) to conduct censuses, but only if we change our approach may we be able to find them.

*       This document was reproduced without formal editing.

**     Central Bureau of Statistics, Israel. The views expressed in the paper are those of the author and do not imply the expression of any opinion on the part of the United Nations Secretariat.