Study on the Continuous Population Census Modality - EMCDC
The Brazilian Institute of Geography and Statistics – IBGE plays the important role of providing Brazil with information that can portray its reality for the exercise of citizenship. This includes the responsibility of carrying out population census. Today the IBGE, as the other Statistical Institutes of several countries all over the world, finds itself in the middle of a dilemma: to produce more and better information in the midst of budget restrictions and urgency of results.
Aiming at improving its work methodology concerning the planning and execution of the population censuses, the IBGE makes an effort to examine and assess methods other than full enumeration of the population, especially those based on samples which are already being adopted in the United States and France.
The project Study on the Continuous Population Census Modality – EMCDC is grounded on the current international discussion on the use of such methods due to the challenge that the National Statistical Offices face to produce updated, geographically and thematically detailed information, with the constant necessity of cost reduction. Knowing that it also must cope with such hindrances, the IBGE has started some activities to foster the study and the assessment of alternative methodologies for the population census in the Brazilian context.
With activities initiated in 2004, the project has been welcoming actions that help evaluate the technical and operational viability of the methodological change. Of note are the three international seminars carried out in partnership with the Instituto Nacional de Estadística Geografía e Informática (INEGI - Mexico), the participations in international symposiums and international meetings on alternative census modalities, including technical visits to the National Institute INSEE (France), besides the participation in the debate on the future of the Brazilian population census in activities at the national level, as CONFEST/CONFEG and ABEP, among others.
Taking part in those activities has helped foster the debate on the subject in the Institute and, in 2004 and 2005, multidisciplinary work groups (WT) with technicians from the several IBGE organizational units were engaged in the project development. Five work groups were created and later defined to encompass the different dimensions of the work, given the project’s complexity:
- WG Census Mapping and Address Register;
- WG Conceptual Design;
- WG Information Sampling, Estimation and Accumulation;
- WG Distribution of Agencies;
- WG Integration between Censuses and Surveys.
Based on the knowledge produced in the different thematic groups a proposal was elaborated for the operational model. It established guidelines for the carrying out of the Pilot Test of the Continuous Census. The test was carried out in the period between 2008 and 2013 and comprises two cumulative cycles: 2008-2012 e 2009-2013.
The preliminary results of the study may be seen in the articles: Study on the Continuous Population Census Modality, presented in the XVIII National Meeting of Population Studies and in the 20th National Symposium on Probability and Statistics; and Rolling Census: scoping study for Brazil, presented in the 59th World Statistics Congress.
The studies must be concluded by the end of 2014, when the implementation limits and possibilities for the Continuous Census Modality in Brazil will be presented.fico Contínuo no Brasil.
About - Operational Model
The proposal of an operational model of census based on an alternative modality to complete enumeration of the population, called Continuous Census was elaborated by technicians of the directorates of Survey, Data Processing and Geosciences of the IBGE. The proposal took into account the discussions taken place between 2004 and 2006, which involved five Work Groups formed by the IBGE and several National Statistics Offices (NSOs), especially Brazil, Mexico, France and the United States, which took part in three International Seminars on alternative census modalities.
The main focus of the proposal is the operational issue; however, technical/methodological aspects were also considered, including information sampling, estimation and accumulation; conceptual design/questionnaire; census mapping and address register; collection network structure and integration with traditional census and other household surveys.
The operational model proposed was based on debates on: which kind of information to produce; the data accumulation cycle of the census operation and on the dissemination periodicity; the sampling model to be used and the geographic coverage allowed by the operation; whether it should be spread throughout the year, or concentrated in a few months as in the regular census; which census mapping the operation should be grounded on; whether the listing of the sectors should be done together with the collection or at different times, and which information should be collected during this operation; and whether the population count is necessary and at which intervals. Moreover, additional questions were raised for reflection, such as new constructions, new municipalities, etc.
The modality of continuous population census assessed in the IBGE consists of a sampling survey with the collection of the information distributed over a determined period of time. In this modality the use of different sampling fractions in groups of municipalities defined by their sizes is proposed. A study was carried out in order to define these groups and the corresponding sampling fractions, in order to guarantee the desired precision of the estimates in all municipalities.
The results found, that is, the groups and fractions, are suitable for any population censuses, since, in order to make estimates with the required quality, a total sample size is needed. It can be obtained either at the same instant, as it is done in a census carried out in the traditional way, or it can be cumulative, if the sample is collected continuously over a period of time, which is the proposed way in the alternative modality studied.
The factors considered for the definition of the groups were:
- Characteristics of interest;
- Accuracy of estimates and
- Sampling fractions or cut limits.
In some alternatives, the fractions were fixed and then the limits were found. They were defined in such a way that the coefficients of variation of the proportion estimates were within the preset limits, for any size of municipality. Thus, for all municipalities the estimates would be obtained with equivalent precision. For other alternatives, the work procedure followed the other way around. The limits were first defined and then the needed fractions were found to estimate the proportions with the desired precision. The sampling fractions proposed, according to the size classes of the municipalities, are shown below.
Table 1 – Sampling fractions by size class of the municipality in total inhabitants
|Size Class of Municipality (inhab)||Sampling Fraction|
|Up to 4,000||50%|
|More than 4,000 up to 8,000||33%|
|More than 8,000 up to 20,000||20%|
|More than 20,000 up to 500,000||10%|
|More than 500,000||5%|
Detailed information on the studies performed for the sample definition and other correlated aspects can be found in the document “Sampling Aspects”.
The studies carried out concerned with identifying the best way to meet the growing demand for information about the characteristics of the population and of the households. The content of the questionnaire and the formulation of the questions were studied. They strongly related to the collection procedures adopted and the planned way of accumulation.
The conclusion drawn was that the subjects to be investigated through an alternative method of Population Census should be those absolutely necessary for the municipal level. The most frequent information demands, essential for municipal management, were identified and assessed. As for the other subjectss, they can be obtained from the other household surveys of the Institution, such as the National Household Sample Survey - PNAD, the Monthly Employment Survey - PME and the Consumer Expenditure Survey - POF.
Information that is not demanded at the municipal level, but is usually provided by the Census, has also been identified, since the sample design of the other household surveys does not allow cross-comparison with a high level of disaggregation. As an example, the National Accounts stand out, which use a level of disaggregation of the economic activities that PNAD does not have.
The work was carried out in subgroups by subject with the objective of analyzing the needs for the changes that an alternative model for population censuses could entail on the different subjects. The following were studied: household characteristics; demographic characteristics; color or race, people with disabilities; education, fertility, infant mortality and migration; characteristics of labor and income; families in the population census and agricultural establishments.
For the selected subjects, the following steps were taken: comparison between the variables investigated in the Brazilian, French and American Censuses; and analysis of the existence of a change in investigation model in the transition between a traditional census and an alternative modality in the two countries under study (United States and France). The purpose of the comparison was to understand how the model change influences the way information is collected for each subject.
In the proposed operational model, the need to meet at least the existing demands was considered, therefore, it was a consensus that the questionnaire to be adopted should contain the questions used in the questionnaire of sample of the Population Census.
Detailed information about the studies carried out for the elaboration of proposed questionnaires can be found in the document “Conceptual Design”.
The proposed periodicity for the dissemination of information to all municipalities is annual, once the first cycle of 5 years is completed. The annual results will be obtained from the five most recent samples. Thus, once the fifth year collection is done, results will be produced based on the samples of the first five years. In the sixth year, the information produced will be based on samples from years 2 to 6, and so on.
In addition to detailed results for the municipal level, based on five samples/years, it is possible to obtain results for the country and large regions based on the one-year sample, as of the first year. With the accumulation of two, three and four annual samples, the geographic level for which it is possible to produce information narrows down.
The results will be referenced in the year in the middle of the cycle, as shown in the diagram below.
Sample i = 1/5 of the total sample. This value corresponds to the sample collection in 1/5 of the sectors of each Brazilian municipality, making up 1/5 of the total Brazilian sectors.
RCi = Result of year i corrected for the middle year of the cycle
The accumulation of RC1, RC2, Sample 3, RC4 and RC5 provides the results for year 3 of the accumulation cycle. In this model, the information produced will be far from the reference date for two years at most.