24 September 2001
Symposium on Global Review of 2000 Round of
Population and Housing Censuses:
Mid-Decade Assessment and Future Prospects
Department of Economic and Social Affairs
United Nations Secretariat
New York, 7-10 August 2001
Census taking in countries with small populations:
Limitations and Potentials*
Frank Eelens **
(Netherlands Interdisciplinary Demographic Institute)
Monique Plaza-Maduro **
(Central Bureau of Statistics - Aruba)
1. At the turn of the millennium, 76 countries had a population of less than one million. Out of this group, 28 had a population between 100,000 and 500,000, while 39 countries had a population of less than 100,000 (United Nations, 1999). These countries vary in terms of social and economic development, population composition and distribution as well as of environmental conditions.
2. The steps for taking a census in a country with a small population are basically the same as in a country with a large population. Through an integrated programme of data collection and compilation, population and housing statistics are generated for economic and social planning. However, the circumstances under which a census of a small population is taken are often quite different from those in larger populations. The mere size of the population sets certain limitations, but it also creates some interesting possibilities. We shall explore some of these limitations and possibilities by elaborating on each of the six issues that the Symposium will focus on. These are:
3. For this paper, we relied on our own experiences concerning the 1991 and 2000 population and housing censuses of Aruba as well as on other censuses in small countries and regions. Aruba has a population of 90,000.
4. Whatever the size of the population, a number of activities remain the same. For instance, the size of a country’s population does not influence the workload involved in the drafting of questionnaires and training manuals, developing the legal frame, computer programming and tabulation. An important characteristic of census taking in small countries is that the entire operation is organized by very few people, sometimes not more than three or four. This means that specialization is almost non-existent. On Aruba, for instance, a census officer might be mapping in the morning, have a meeting with a publicity agent in the early afternoon and put the finishing touches to editing programs later that day.
5. Having a small staff has a management advantage. Each census officer has a much better overview of the operation. Possible flaws and weaknesses in the census operation and data might be more easily recognizable and dealt with. On the other hand, census officers are no specialists in each aspect of the operation. This fact might bring the quality of each aspect in which the officer is involved into question. On Aruba, this risk was reduced by bringing in external specialists for short periods to assist in those fields where internal expertise was lacking. This was the case in the field of automatic data processing and, to a lesser extent, in the case of mapping.
6. Another option is that a group of countries joins forces and rounds up its specific expertise. During the 1990 round of censuses, a group of countries of the Caribbean Community (CARICOM) centrally coordinated the preparation and processing of its censuses (UN ECLAC/CDDC, 1998). In the current round, the same countries are following this approach once again. There is no doubt that members benefit from such collaboration. However, a centralized approach between a group of countries certainly involves some risks. Red tape and financial constraints may affect the overall preparation; a delay in one country might slow down operations in all the others. For instance, if scanning equipment is shared, a deceleration in the processing in one country has a direct effect on the timing of data processing in the others.
7. Many small countries have only limited research capacities. This means that social and economic surveys are held only periodically. Many organizations consider a census a golden opportunity to provide answers to numerous questions. On Aruba, a staff member of a department proposed to include 30 specific questions for the 1991 census. All his questions were valid and well thought-out. This man was therefore most disappointed when only two of his questions made it into the questionnaire.
8. A census is often the only source of valuable information in a small country, and this might lead to serious pressure on the census staff to include certain questions in the questionnaire. The fact that professional, political and social networks are small may further increase this pressure.
9. Census users need not only direct results from a census, but also findings and recommendations based on the research of census data. Universities, governmental and private research institutes and non-governmental organizations (NGOs) take on this role in larger countries, but second-line research institutes are generally absent in small countries. That is why census data in small countries may be underutilized.
10. An important aspect of the relationship between the census office and the stakeholders is the supply of timely, reliable and appropriate data. In this, smaller countries have a possible advantage over larger countries. Because of the limited number of forms, census results can theoretically be made public relatively soon after the census is held. On Aruba, the population and housing census was held 14-21 October 2000. A CD-ROM with 110 selected final tables of the census was presented to the public in June 2001—eight months after the census week.
11. Small countries have the potential to have better data quality control than larger countries. During the census on Aruba, census staffers made daily visits to each of the 55 enumeration centers. Supervisors and controllers in the field checked every questionnaire for errors. Census staffers could give direct advice when problems arose and they could get instant feedback on quality issues. In large countries, census staffers can only dream of such a direct link between the census office and the individual enumerator.
12. Small countries have another advantage in terms of data quality control during the editing phase. A large and often underestimated quantity of errors is introduced during the data-entry and coding stages. Even with optical mark reading and optical character recognition, more than just a few errors are generally introduced. In small countries, it is much easier to go back to the census forms and make corrections. Larger countries have to rely on automatic imputations for that. On Aruba, census staffers had the forms beside them while editing the data from an enumeration area. There is no doubt that this has increased the quality of the final data. Obviously one should not overdo the editing and keep a healthy balance between accuracy and too much time-consuming correction. There might be a tendency to treat a census as an accounting operation instead of a counting operation in a small country.
13. The most efficient and effective source of statistics is not directly linked to the size of a country’s population, but more to the availability and quality of alternative sources. Current availability of information from administrative sources and surveys would certainly not allow for abandoning the census on Aruba. As in many small countries, the overall research capability on Aruba is rather limited in terms of human resources and financial means.
14. Again, because of the size of the population and consequently of the overall control over the census operation, some other topics could be incorporated within the census. These topics require a somewhat more sophisticated approach and would normally be addressed in surveys in larger countries. For instance, by proper indexing persons in family nuclei and in households, the census on Aruba provided a very detailed classification of households by type. This was done according to the number of family nuclei and the relationship between the family nuclei and the other members of the household. Moreover, the information provided a detailed classification between conjugal and consensual family nuclei.
15. New high-technology techniques should be seriously considered only when their advantages are abundantly clear and the associated risk of failure is sufficiently small to be acceptable. Census planners should be aware of internationally proven technologies and concepts. Equally important, they have to establish the most appropriate techniques to gather and process the required information (Dekker, 1996). Planners have to decide at the start of the census which technologies are appropriate for their country, given the size of the population, the available knowledge and skills and financial resources. In recent years, significant progress has been made in the field of scanning and imaging. Most countries that have adopted automatic data entry use a combination of Optical Mark Reading (OMR) and Optical Character Recognition (OCR). In some cases imaging is used to scan text boxes with written information on education, profession, kind of industry and so forth. If used properly, optical data entry leads to a reduction in processing time and cost.
16. If used solely for data entry of a population census, optical methods are often not the right choice for small countries for the following reasons:
17. Should scanning technology for census processing be introduced in small countries, then it should form part of a larger plan to use this technology in other statistical fields as well. In the 2001 population and housing census on Aruba, it was decided to use OMR/OCR and imaging. This technology is now also used for other purposes, e.g., the monthly tourism expenditure surveys. By using optical data entry combined with computer-aided coding, total processing time of the census was four months shorter than the previous census.
18. The decision to work with optical data entry was influenced by several important considerations. Aruba has a very tight labour market. Good data-entry typists are hard to find and salaries are relatively high. Although local experience with digital data entry was limited, expertise from abroad was available through a cooperation project with the Netherlands Interdisciplinary Demographic Institute (NIDI). Expertise with sophisticated questionnaire design and form printing was available in-house.
19. Transforming verbal answers into numerical codes is a major, time-consuming activity in most censuses. In the past, efforts have been made to use computer technology to facilitate this job, either through automatic coding or through computer-assisted coding, but any use of automatic or computer-assisted coding has to be critically examined. As Dekker (1996: 1137) states: “One must remain skeptical about the application of either computer-assisted coding or automatic coding in developing countries. Only where difficult-to-code variables are absent or hierarchical coding schemes for such variables remain shallow, can the case for computer coding be made”.
20. Smaller countries definitely have an advantage over larger countries in this field. Most of these countries’ economies are far less complex. Many have economies that are based on two or three main pillars. Consequently, the difficult-to-code variables are more limited. For instance, it is much easier to construct a look-up table for the variable “industry” in a small country. The Central Bureau of Statistics (CBS) in Aruba has been keeping a register with all local businesses for years. This file contains the International Standard Industrial Classification (ISIC) code for each establishment and can easily be linked to the system of computer-assisted coding. On Aruba, the system was an integral part of the verification stage of the optical data capturing. Scanned boxes with written information were projected on the screen of the verifier. The verifier first typed in a few characters of the written description provided by the enumerator. The system then responded with a list of possible codes to choose from.
21. Census results are traditionally distributed through printed reports. A special report is made of a group of general tables that have been selected and grouped. Since the 1990 round of censuses, technology has further developed and more alternatives are now available for dissemination of census results. The CD-ROM as a publication medium has many advantages, especially for small countries, because:
22. A population and housing census is the largest administrative undertaking in a country. Censuses have many spin-offs for statistical offices. They always provide a great learning experience for all those involved. Knowledge of administrative, verifying, coding and editing procedures acquired during the census are a great asset for further research activities. Especially in small statistical offices, where a large proportion of the staff is involved in the project, a census may have a profound effect on its institutional development.
23. It is important for future reference that all census material is stored, both physically and in digital format. All letters, notes, computer programs, training manuals, coding lists, tables and reports should be put on CD-ROMs. Experience shows that it is definitely worthwhile to make an index of all materials stored. Ten years is a long time to remember cryptic file names!
24. Systems used during the census should be updated on a regular basis, the most important being coding lists and census maps. In small countries, updating coding lists can be done on a regular basis through surveys. On Aruba, new occupation and industry names, together with their respective codes, are added to the list whenever a labour force survey is held. At the next census, these coding systems are ready for use and priorities can be given to other issues. As we stated before, small countries do not have such diversified markets as large countries; this facilitates updating code lists.
25. The data contents of a population census and their utility in the preparation of development plans will depend largely on the consistency and the compatibility of the definitions and concepts used (Maitra, 1996). In a small country like Aruba, the same staff is involved in the preparation of a census as well as in the coordination of survey research. This facilitates the continuity of use of concepts and definitions during the intercensal period.
26. At the onset of the census project on Aruba, no maps were available. No other department had updated digital maps. Therefore, the CBS decided to develop its own census cartography, which resulted in one of the most extensive projects of the Aruba Census 2000. In the period July-October 1999, CBS staffers completed a first round of fieldwork. The information gathered in the field was used to update the digital maps created by the Department of Land, Surveying and Land Registration in 1990 and the aerial photographs taken in 1998. Additional information on new housing projects and roads was provided by the Department of Public Works. Four months prior to the census a second round of fieldwork to update census maps was completed. Based on these digital maps and the information gathered in the field, a Geographical Information System (GIS) was created. Census enumeration areas and districts were added as different layers to the GIS. The GIS played a key role during the census planning, the field operation, the data processing and the final analysis. Little by little, this system grew into a full-fledged geographical information system. Our GIS is now drawing a lot of attention from other governmental departments as well as from the private sector.
27. Not many small countries in the Caribbean have been able to produce their own GIS. They have not been able to fully update their census mapping for the 2000 round of censuses. The Aruba experience teaches that it is possible for a small statistical office to develop its own GIS system. To be successful, however, one needs to stick to some basic principles:
28. There are numerous methods to estimate the coverage of censuses. Demographic analysis, administrative sources and post-enumeration surveys may all be used to examine the thoroughness of a census. The method or combination of methods that works best in a particular country depends on the availability of information and resources and is independent of the country’s size or population. On Aruba, the census team mainly used administrative sources to estimate the degree of non-coverage and special forms to monitor progress during fieldwork.
29. A population census is always an adventure for those involved. Censuses in large countries are normally executed by a large staff. The workload is divided along regional and topical lines. In small countries, census planners have to be able to cover all the diverse aspects of a census with just a small staff. This forces the staff to be more generalists than specialists. As we saw, this may be a risk for the quality of the census operation.
30. Ten years is a long time in any organization. In that time, people are promoted, retire, find other jobs or simply think that one census was enough. That is why a census team often has new members—people who have never been involved in a census before. Consequently, certain skills might be lacking at the onset of a census operation, but these problems might be solved by collaboration, external technical assistance and training. Training is the most obvious of all: changing team members into specialists in the various aspects of census taking will increase the overall quality of the project. Because of the fact that specific know-how might not be available in small countries, training—sometimes abroad—has to be provided. For specific lack of expertise, assistance from abroad might be the answer. For instance, the introduction of new technology may require a foreign expert. Finally, collaboration between countries is an ideal means to overcome weaknesses in each of the individual members. The collaboration between the member states of CARICOM to coordinate their efforts is a typical example of how to overcome problems created by size.
31. In some ways, small size is a problem in censuses, but in other ways, it is a blessing. Throughout the various stages of the census, easy quality control can take place in a small operation. Coding becomes less of a quality risk and editing can be done more profoundly. Training of field staff can be given by the census team itself and field control can be more intense. Finally, the sense of ownership of the project is tremendous. There is no better feeling than to get together with your team members at the end of a successful enumeration and look back on the positive impact that each of your colleagues has had on every aspect of the entire operation.
Dekker, A. (1996). Innovative computer methods for demographic enquiries and the dissemination of population information. In Proceedings of the Expert Group Meeting on Innovative Techniques for Population Censuses and Large-Scale Demographic Surveys, 22-26 April, The Hague.
Maitra, A. (1996). Population censuses and surveys in the next century. In Proceedings of the Expert Group Meeting on Innovative Techniques for Population Censuses and Large-Scale Demographic Surveys, 22-26 April, The Hague.
United Nations (1998). Principles and Recommendations for Population and Housing Censuses. Revision I. Statistical Papers, Series M., No. 67/Rev. 1. Sales No. E.98.XVII.8.
United Nations ECLAC/CDCC (1998). Report on the Workshop on the Year 2000 Round of Population and Housing Censuses, 2-4 December, Port-of-Spain, Trinidad and Tobago.
United Nations Population Division, Department of Economic and Social Affairs (1999). World Population Prospects: the 1998 Revision, Vol. I: Comprehensive Tables. Sales No. E.99.XIII.9.