NIGERIA’S EXPERIENCE IN CARRYING OUT A POST ENUMERATION SURVEY (PES): IMPLICATIONS FOR CENSUS COST SAVING STRATEGIES.
Dr. Akintobi A. Kadejo
Director General
National Population Commission
Federal Republic of Nigeria
International Expert Group Meeting on Mechanisms for Ensuring Continuity of
10-Year Population Censuses: Strategies for Reducing Census Costs.
Pretoria, South Africa, 26-29 November 2001.
NIGERIA’S EXPERIENCE IN CARRYING OUT A POST ENUMERATION SURVEY (PES): IMPLICATIONS FOR CENSUS COST SAVING STRATEGIES.
Dr. Akintobi A. Kadejo
Director General
National Population Commission
Federal Republic of Nigeria
Introduction
The population census is the main source of information on the number and important demographic, social and economic characteristics of the population in a country. The uses of census data are innumerable. They are used in formulating policies for national development such as manpower development, formulation of housing policies and programmes, planning for social welfare programmes, and formulation of policies for investment (United Nations,1992).
A population census is also the most elaborate and costly data collection activity in a country. Its distinctiveness such as completeness of coverage, universality, simultaneity, and periodicity not only makes a census a complex exercise but also a costly operation. While three activities take-up the bulk of census operation cost, (a. census mapping and Enumeration Area Demarcation (EAD), b. population enumeration, and c. data capture, processing and dissemination), other factors like continuing high rates of population growth, high illiteracy, the lack of an efficient civil registration system, and extensive population movements, have contributed to the spiraling cost of censuses, especially in Sub-Saharan Africa.
It is becoming increasingly more and more difficult for countries in Sub-Saharan Africa to secure funding for censuses, due to economic and financial crises in most countries, and bilateral and multilateral donor fatigue. Hence, it has become imperative to examine cost-saving strategies in undertaking censuses.
PES is an important component of census evaluation, and under the right conditions, the benefits could be substantial. Many countries have used PES successfully in evaluating their data, and there are several examples of best practices and the documented usefulness of PES (Whitford and Banda, 2001)
This paper has two objectives: first, to discuss briefly Nigeria’s experience with PES; and second, to examine its implications for cost-saving strategies for census operations.
A. Post Enumeration Survey in 1991 Nigeria Population Census.
It is a good census practice to evaluate the completeness and accuracy of the results. The UN, therefore, urges that a post enumeration survey must be an essential part of the overall census operation. In this regard, the Nigeria Population Census program of 1991 had a PES as one of its major activities. It was designed to serve as an instrument to cross-check the accuracy or otherwise of the census results.
Also, the objective was to collect additional information especially on fertility, mortality and migration. The PES collected information on current and retrospective fertility from women 10 years and above. In addition, children ever born, children surviving, birth and deaths in the last year, were canvassed together with migration information.
The additional topics were:
(i) Duration of residence and LGA (Local Government Area) of previous residence;
(ii) Educational attainment;
(iii) Employment status;
(iv) Age at first marriage;
(v) Children ever born;
(vi) Children alive;
(vii) Births in the last 12 months;
(viii) Births in the last 12 months, alive; and
(ix) Number of deaths in the household during the last 12 months.
This strategy was aimed at putting minimum questions in the main census questionnaire so as to enhance completion of enumeration within the census period and reduce respondent fatigue.
Sample Size:
The PES was a two-stage sample design that covered 5% of the delineated Enumeration Areas (EAs). It was a random sample design using the census enumeration areas frame. The states in Nigeria were used as the first stage of the stratification and there was independent sample selection from among the states. Additional gains from stratification were achieved from systematic selection of sample EAs from geographically ordered lists of both the Local Government Areas and the census EAs.
The five percent sample used was considered parsimonious to meet the conceived objectives. A large sample would have involved more resources and longer enumeration time. The allocation of the samples among the 31 states was arrived at in order to maximize precision at higher geographic levels.
Implementation of The PES
The PES was conducted three weeks after the 1991 Census throughout the country. A detailed calendar of operations was used during the implementation. Immediately after the census, a refresher course was organized for the best supervisors and enumerators used for the census. These functionaries were deployed to the sampled areas not necessarily where they worked during the census. The training and PES materials distribution took place between 13th -14th December 1991. Some sampled EAs were identified, and housing and building numbering carried out from 17th -19th December 1991. The last phase was the enumeration of the respondents and retrieval of documents from 20th - 23rd December 1991.
After the retrieval of materials, all the follow-up activities, manual editing, data capture, secondary editing and tabulation followed the same pattern as the main census.
The PES was expected to be followed by a matching exercise, comparing census and PES questions and data sets on a case by case basis, with the objective of identifying missed, duplicated and/or fabricated persons. Thus, for effectiveness, the Commission developed adequate and appropriate matching rules.
However, after the preliminary analysis of the Census/PES results, it became obvious that the PES results could not be effectively used to evaluate the census as planned. A further study of the result suggested that the PES was conducted under a completely different condition from the census.
Firstly, compared with the main census when there was a three-day restriction of movement, there was no restriction of movement during PES.
Secondly, the PES was conducted during the Christmas time when many families had started to return to their home states for the festival.
Thirdly, many people were apathetic to answering the numerous questions.
All these accounted for the high rate of call-backs and “refused interview” at the end of the exercise.
Therefore, the PES was not used to estimate coverage error. However, all the additional information on fertility, mortality and migration collected through the PES were adequately utilized in the analysis of the census results.
From the experience of the 1991 Census, future PES should be designed to avoid all the past pitfalls and constraints, so that it will be worth the cost. There is need to find pragmatic and lasting solutions to them.
Strategies for cost- saving in Future Censuses
The 1991 PES did not have any preceding PES as a point of reference due to the history of census taking in Nigeria. Nevertheless, based on the 1991 experiences, future PES should adopt the following strategies in order to achieve the objectives especially for evaluation of the census results:
1. The basic cost-saving strategy is to use an efficient sample size (e.g. 1%, 5 % or 10%), and sampling strategy. An optimum allocation procedure should be maintained. What you allocate to the State should be with Probability Proportional to Size (PPS) and it should be proportional to variability within the stratum. The larger the variability, the larger the sample size allocated to the cluster and it is inversely proportional to the cost - the unit cost of deriving information. Conversely, the higher the cost the smaller the sample size allocated.
2. Absolute minimum core questions should be used. Using more questions would take longer time and this will require more man-power and materials that would bloat the budget of the PES. If absolute minimum core questions are used, the cost of PES can be considerably reduced.
3. PES should be treated as a continuous exercise with the same tempo as the main census, the IEC/Advocacy machinery for the census will include information on PES from the outset to inform the respondents of possible second interview if they fall within the sampled EAs. There will therefore be no need to include an extensive budget for publicity in the PES. This option becomes more attractive if there is no much time lag between the census and the PES. The census messages, which also contain information on PES, will still be fresh in the memory of the people.
4. There should be an unambiguous PES plan stating the various requisite activities as part of the overall census program . A separate team should be in charge of the PES activities from its inception. That is, from methodology, sampling, enumeration and post-enumeration activities with its own budget. This arrangement would ensure that the PES is not overwhelmed by the enormity of census preparatory activities. Under this scheme, the PES would be given the necessary attention and resources required for its smooth conduct. However, there must be strong and efficient coordination between the main census and the PES.
5. The ideal time for conducting a PES should be the immediate period following the census but not later than two months. This becomes desirable if there is not going to be restriction of movement like in the census.
6. PES methodology from sampling, through processing, including the matching which generally constitutes the bottle-neck and the calculation of variance should be adequately computerized. In the case of Nigeria, it is anticipated that scanning techniques (Optical Character Recognition/Optical Mark Reader) will be applied for the main census. It is recommended for cost effectiveness the same technique should be used for the PES and the matching exercise should be computerized as much as possible even if it requires a specific program to be developed locally. However, if any standard packages are available for the purpose, it is encouraged to use them. The data processing of PES should be included in the work plan to ensure prompt tabulation. However, processing of PES should be kept different from the processing of the census, otherwise we may not get the results of the PES. This is also a cost saving strategy. It means that the same data capturing facilities and programs used for census can also be used for the processing of the PES data.
In conclusion, all these strategies are meant to ensure that we obtain reliable and cost effective census and PES results. Not getting results means that the PES is just an expensive exercise in futility.
References
National Population Commission (1998): 1991 Population Census of the Federal Republic of Nigeria Analytical Report at the National level, Abuja: National Population Commission.
United Nations (1998): Principles and Recommendations for Population and Housing Census (Revision 1), New York.
U.S Bureau of the Census (1985): Evaluating Censuses of Population and Housing, Washington, D.C., (1985).
United Nations (2000): Handbook on Census Management for Population and Housing Censuses, New York.
United Nations (2000): Handbook of Population and Housing Censuses, Parts I and II, New York
Whitford, D. C. and Jeremiah P. Banda (2001): “Post Enumeration Survey (PES): Are They Worth It or Not? In Symposium on Global Review of 2000 Round of Population and Housing Censuses: Mid-Decade Assessment and Future Prospects, Statistics Division, UNDESA, New York, 7-10 August 2001
7