A. Introduction

1. The 2000 round of censuses in Latin America has encountered several challenges, some of them new and others long-standing. In the early 1990s, all signs pointed to an economic upturn, hence, to a reduction in fiscal pressures. In the last few years, however, the region was plagued with major economic difficulties and with sharp cutbacks in fiscal spending. This, clearly, affects the censuses, which are costly operations requiring timely financing to carry out each programmed phase and ensure accurate results.

2. Whether the content of the censuses is relevant to the problems currently affecting the countries of the region is another issue to be resolved. Reference is made to the crucial social issues—for example, poverty and inequities, the environment and health and social security reforms.

3. The first section of this study reviews some aspects of the census tradition in the region, and the second section explains the role of population and housing censuses within the national statistical system. The third section refers to some of the financial aspects of the censuses and alternative arrangements for defraying related costs. The next two sections deal with conceptual and technological innovations which have been discussed or incorporated, while the last section analyses methods of evaluation of the results.

4. The information and ideas contained herein were obtained primarily through contact with the countries through technical assistance, seminars organized in the countries and at the headquarters of the United Nations Economic Commission for Latin America and the Caribbean (ECLAC) (ECLAC, 1999), as well as from a survey held in 1998 of the statistical institutes and offices in the region on the experience of the last census (Del Popolo, 1999).

1. The census tradition

5. Although censuses were held in several Latin American countries for different purposes in the first half of the 1900s and even earlier, it was not until 1950 that population and housing censuses started to be held systematically in most of the countries of the region. Beginning with this decade, countries have attempted, with varying degrees of success, to follow the recommendations of the United Nations by holding such censuses once every ten years, where possible in years ending in zero, and by ensuring that they are conducted simultaneously throughout the national territory. Of the 20 countries in the region, only one did not conduct a census around 1950, and fifteen did so in that very year. Since then, the tradition has continued, although always with some exceptions, since, for financial or political reasons, some countries have had to skip a decade or to hold their census after the year ending in zero. For example, in the 1990s, Costa Rica and Cuba did not hold censuses, Honduras held one in 1988 and of the total of 20 countries, only four held censuses in 1990. In the current decade, a similar situation has arisen: censuses were held in 2000 in only four countries: Brazil, Costa Rica, Mexico and Panama; five other countries are planning to hold their censuses in 2001, while more than half of the remaining countries are scheduling theirs for the following years.

6. According to joint evaluations made by national agencies and the Population Division, Latin American and Caribbean Demographic Centre (CELADE), census coverage is considered generally satisfactory. Except for a few cases, the percentage of omission has been less than 10 per cent and, in a considerable number of cases, less than 5 per cent (Table 1). The interesting point to note is that, over time, there has not been any clear improvement in population coverage. Instead, in some cases, there has actually been a decline. This may be due to a number of factors, such as the increasingly complex situation arising from high population growth, especially in cities, improved techniques for evaluating the quality of information, the difficulty in incorporating new concepts and more sophisticated technology.

Table 1

Latin America: Percentage of census omission*

Censuses from 1950 to 1990

Country	Census round
Country	1950	1960	1970	1980	1990

Argentina	1.4	3.3	2.8	1.0	0.9
Bolivia	0.7	-	5.9	-	6.7
Brazil	3.8	4.2	3.4	2.5	2.5
Colombia	12.9	8.1	19.1	12.5	11.2
Costa Rica	6.7	2.2	-0.4	5.4	-
Cuba	5.0	-	-0.3	0.8	-
Chile	6.1	4.0	6.1	1.2	1.1
Dominican Republic	9.5	6.0	8.2	6.1	5.1
Ecuador	6.5	6.1	2.6	5.1	6.8
El Salvador	4.7	5.1	3.8	-	4.4
Guatemala	5.5	2.8	8.7	12.8	13.8
Haiti	5.2	-	6.1	12.0	-
Honduras	0.7	3.2	8.3	7.2	-
Mexico	6.8	5.3	3.4	0.9	1.8
Nicaragua	6.5	9.3	13.8	-	1.0
Panama	13.0	5.7	4.8	5.8	2.6
Paraguay	11.4	6.6	4.6	8.4	7.1
Peru	-	3.1	2.7	4.2	3.0
Uruguay	-	1.7	1.4	2.0	2.3
Venezuela	2.8	3.1	4.5	6.9	7.8

* The omission is calculated by comparing the population enumerated with the estimated population based on population estimates and projections existing in CELADE. Updated in November 2000.

7. The first part of the questionnaires includes the geographic location of the housing unit and questions relating to the quality, occupancy status and basic services thereof. In most cases, each household within a housing unit is counted as a single enumeration unit (Table 2), although some countries fill out only one questionnaire per housing unit and identify the households or families living there. Households are defined as consumer units insofar as their members share the budget and may or may not be relatives. As consumer units, households are asked to respond to a series of questions referring to the number and use of rooms, sanitary conditions, equipment (radio, television, washing machine, telephone, car, etc.). Other questions usually included for investigating specific facts concern the language spoken (to identify indigenous communities), deaths that have occurred in the household (to assess adult mortality), and persons living abroad (to measure international emigration), to name a few. The quality of the housing, the availability of basic services in the housing unit and the items of household equipment are details frequently used for assessing the social stratification of households through the procedure of indicators of unmet basic needs.

8. Once the persons belonging to the household have been listed, questions are included for each individual, following identification of the head of the household. Whatever the format of the questionnaire (a form with persons listed by column or, as is more common today, a booklet listing persons on separate leaves), normally the questionnaire is organized by sections moving from the general to the particular: questions addressed to all persons (relationship to the head of household, sex, age, marital status, residence, and, in some cases, nationality and religion), questions on educational background (addressed to persons over 5 years old or some other similar age), questions relating to economic characteristics (addressed to persons over 15 years of age or, in some cases, younger), and lastly questions addressed to women 15 years and over on the number of children ever born and surviving.

9. In order to decide which persons should be included in each household, the type of count used is important—that is, whether it is a de facto or a de jure census. In the former case, all the persons present in the household on the night prior to the census are included in the count, while in the de jure type, the persons who usually reside in the household are counted, including absent residents and excluding non-residents who happen to be present. The de facto census was more common in the past, for practical reasons, since the concept is clear and easy to understand. However, by 1990, about half of the countries had opted for a de jure census (Table 2), a decision based on conceptual reasons, since it is assumed that this type of census is a more accurate reflection of the true situation and facilitates the study of migration and family composition. Practical arguments have also been advanced to the effect that it would be easier to use the de jure type of census in cases where the field operation lasts several days.

10. Some of the heavily populated, geographically vast countries have used sampling for investigating a series of issues. That is to say, they have adopted a basic universal questionnaire with just a few questions and a more complete form for household samples. The sampling technology seems suitable for certain types of study, bearing in mind its potential for reducing costs and improving the quality of responses. However, this arrangement is not widely accepted in Latin America (Table 2), probably for two reasons. The first, and perhaps the more important, is that decentralization policies at geographically disaggregated levels require representative information in the local sphere, which, in most cases, cannot be obtained through surveys. The census is practically the only source of information for small populations. The second reason is that the use of samples tends to make fieldwork and data processing more complex, since two questionnaires are involved, only one of which is applicable in some households. Moreover, expansion factors must be used in giving the results and these usually complicate the data-processing exercise. Nevertheless, countries with larger populations have incorporated a complete questionnaire for large samples, as was the case in Brazil and Mexico in the census taken in 2000.

Table 2

General characteristics of the last censuses carried out by Latin American countries

Country		Date last census			Date next census			Type of		Unit of		Use of		Number of questions (d)				Survey of
			Month	Year		Month	Year		Census		Enumeration		Sampling Frame		Housing	Household	Persons		Coverage
	Argentina		5	1991		10	2001		De Facto		Household		X		9	7	28		X
	Bolivia		6	1992		9	2001		De Facto		Household				18		23		X
	Brazil		9	1991					De Jure		Housing (c)		X		27		61		X
						08	2000 (a)		De Jure		Housing (c)		X		23		67		X
	Chile		4	1992			2002		De Facto		Household				10	6	19
	Colombia		10	1993			2002		De Jure		Household				6	7	21		X
	Costa Rica		6	1984					De Jure		Household				13		18		n/d
						6-7	2000 (a)		De Jure		Household				16		20
	Cuba		9	1981			2002		De Jure		Household				14		19		X
	Dominican Rep.		9	1993		11	2001		De Facto		Household				6	5	23		n/d
	Ecuador		11	1990		11	2001		De Facto		Household				10	8	23		X
	El Salvador		9	1992			2003		De Jure		Housing				21		28		X
	Guatemala		4	1994			2004		De Jure		Household				10	12	24		X
	Haiti		9	1982			2001		De Jure		Household				5	8	27		X
	Honduras		5	1988			2001		De Jure		Housing				16		30		X
	Mexico (b)		3	1990					De Jure		Housing		X		18	6	33
						2	2000 (a)		De Jure		Household		X		20	12 (e)	37
	Nicaragua		4	1995			2005		De Jure		Housing				13		25		X
	Panama		5	1990					De Facto		Housing				13		25
						5	2000 (a)		De Jure		Household				15	21	27		n/d
	Paraguay		8	1992		8	2002		De Facto		Household				12	12	21		X
	Peru		7	1993			2002 or 2003		De Facto		Household				9	5	23		X
	Uruguay		5	1996					De Facto		Household				11	11	31
	Venezuela		10	1990		10 or 11	2001		De Jure		Household		X		23	12	26		X

Notes: n/d: no data. (a) Censuses already conducted. (b) In the case of Mexico the information relates to the Population Count held in 1995. In the enumeration, the unit of enumeration was the housing unit but a sample survey, which was held jointly, identified households. (c) In Portuguese, the term “domicilio” is used to mean housing. (d) In the case of sampling frame reference is made to the expanded questionnaire. (e) It contains an International Migration Module.

11. In order to assess the coverage and quality of the census, the use of post-enumeration surveys is very widespread, although there is consensus that they pose a number of problems in the field and at the level of data analysis. In any event, in most cases, the direct evaluation complements the meticulous study of census data and collateral information through indirect methods.

2. The census: an important component of the national statistical system

12. The modern view of the population and housing census is that, in conjunction with administrative registers and continuous and special sample surveys, it forms an integral part of the national statistical system. In addition to providing basic information for socio-demographic assessments and inputs for decision-making, the census provides data on the target population for social policies and programmes. Population and subpopulation data are used to calculate rates of every kind and per capita averages. It is interesting to observe that in some countries it is proposed that the census, which is a costly operation, should be considered not as an expenditure but rather as a basic investment for maintaining an up-to-date statistical system.

13. As it is a massive and complex operation, the census comprises a limited number of questions, which are usually straightforward and short so that the interview does not take up too much time and the concepts are easy to grasp. Nevertheless, since they are universal, they constitute the ideal framework for obtaining samples which enable them subsequently to carry out in-depth studies on the most diverse issues and for the subpopulation that is to be identified. Thus, in Latin America, practically all countries have a continuous household survey for multiple purposes, although, in some cases, it is not national in scope. Currently, a number of countries subscribe to the Programme of Technical Assistance for Improving the Surveys of Living Conditions in Latin America and the Caribbean (MECOVI), which has the support of the Inter-American Development Bank (IDB) and the World Bank.

14. In many countries, the population census is used as the basis for establishing proportional representation in Parliament and as an instrument for allocating central government funds to the municipalities. The latter has even been cause for concern among some statistical offices, which fear that census figures could be manipulated or attempts made to influence population estimates and projections with a view to favouring a certain area in particular.

15. In addition to considering the census as the basis for the national statistical system, there is also an appreciation of the importance of coordinating censuses taken within regions for economic integration. Thus, the Mercosur countries, with Chile and Bolivia, have pioneered an attempt to obtain a minimum number of comparable variables, create a common database and share experiences and mutual support for their own censuses. These countries maintained a project which has enabled them to organize workshops and working groups for coordination of the census. In addition, meetings were held among countries of the Andean group for sharing experience in this area.

16. In addition to providing the information flow for the work of the public sector, the national statistical system should also consider different types of users: universities, research centres, non-governmental organizations (NGOs) and the private firms. Even though modern technology has facilitated information delivery, there is still resistance to this innovation, mainly in microdata delivery.

3. Funding: a key issue in the preparatory phase of the census

17. As in other mass operations undertaken by nations, the cost of the whole census process has significant financial implications. Clearly, the preparation of the census budget and the subsequent procurement of funds are an essential part of the preparatory phase: without funding, the census cannot take place and with insufficient funding, the quality and timeliness of the data are affected.

18. Beyond the problem of inflation and demographic growth itself, censuses today cost more than in the past. This is not due to any negative factor; on the contrary, the current cost reflects the growing demand for information and, above all, the increased use of more specialized or more detailed information. On the one hand, decentralization policies in state administration have led to a greater demand for data at the regional and local levels, and, on the other, the more thorough investigation of certain issues is arousing greater interest among private-sector bodies, whether they be non-governmental organizations, universities or firms involved in production. Part of this greater demand is associated with the extraordinary technological progress which provides rapid access to vast quantities of information which, just a few decades ago, it would have been impossible to generate.

19. How much do censuses cost? This is not an easy question to answer and comparisons between countries are even less straightforward, since there are widely varying criteria to establish the budgetary component of this operation. For example, some budgets include direct and indirect costs; in others the indirect costs are not included; in certain cases, the time devoted to the census by regular staff is taken into account, in others, it is not, and so forth. Even so, we still have certain references that show that census costs vary significantly from one country to another, with budgets that range between US$1 and US$5 per person. The average for the region was probably between US$2 and US$3 per person. Although the total for the 2000 round of censuses runs into millions of dollars (close to US$10 million in Paraguay and Bolivia and approximately US$100 million in Argentina), censuses in Latin America still cost significantly less than those in the developed countries. In the United States, for example, the 2000 census cost over US$20 per person, exceeding US$6 billion. If we use a rough estimate of US$2 per person on average, in the case of Latin America and the Caribbean, where there are over 500 million inhabitants, this implies a total investment of approximately US$1 billion.

20. If we think of censuses as short-term exercises, they may seem excessively expensive. The outlook is quite different, however, if we consider the approximately 10-year period they cover, bearing in mind that the data from one census are used continuously right up to the time of the following census. This operation should be viewed as forming part of a national statistical system, which provides the contextual and sampling frame for 10 years, in addition to being the main universal source that ensures that this system incorporates data relating to small areas.

21. Since the census is part of development programming, which involves the participation of various sectors, it seems reasonable that the cost should be imputed to the statistical system on which decisions relating to such programmes are based. In this regard, the census must be seen as an essential part of the statistical activities that may serve as an input in development programmes and not as an operation which takes a couple of years, even though the expenses are concentrated within a short period.

22. The costly censuses of the 2000 round are being implemented at a time when serious budgetary constraints are affecting national public expenditure owing to the economic situation and structural adjustment policies. This has meant sharp cutbacks in budgetary allocations for this item in some countries in the region. In addition, non-refundable external funds from international agencies or governments of developed countries have also tended to decline in comparison with earlier decades. This shortage of national and international funds has already led some countries to postpone the census and, in some cases, is seriously jeopardizing the chances of its ever taking place.

23. This concern has already been voiced at workshops, seminars and conferences on the issue, where emphasis has been placed on the need to seek new sources of funding, and on resorting to non-traditional sources. At the national level, the involvement of other public-sector institutions as well as the private sector has been put forward as a possible alternative to be considered in greater depth. This is consistent with our view regarding the growing demand for information from different national and local agencies as well as from private businesses and the fact that censuses should be considered as the basis for the national statistical system. Cooperation may not necessarily be expressed in monetary terms, but agreements could be reached for logistical support in terms of allocation of premises, vehicles, staff and other resources.

24. The participation of the private sector, as mentioned by Schkolnik and Guzmán (1999), may be given in various instances: in the use of the results, the implementation of certain census activities, and in the joint financing of some stages of the census. Some hold the view that the private sector benefits from information without assuming even part of the cost of the operation. While some countries do have ways of charging for the information, there is less experience in joint financing of part of the data-collection phases. In some developed countries, for example, service agencies have been established which purchase the census and survey data and transform them into value-added products. This line of association is promising, but there are some legal hurdles including the problem of confidentiality of information.

25. At the international level, consideration has been given to the possibility of resorting to bank loans. This seems a feasible option, considering that the Inter-American Development Bank (IDB) and the World Bank have a programme of support for countries for the Programme of Technical Assistance for Improving the Surveys of Living Conditions (MECOVI). There are also earlier projects, financed by one or the other of these banks, for statistical development, some national in scope and some covering particular sectors. On being apprised of the promising contacts made by CELADE, the Meeting of Directors of Statistics, held at ECLAC headquarters in Santiago, Chile, in March 1999, welcomed this initiative and stated in its resolution that contacts should be maintained with IDB to indicate to that organization the importance of opening up a line of credit to support the 2000 round of censuses.

26. IDB welcomed the idea and several countries have made loan applications. Bolivia, the first country to make such an application, already has one project under way with the Bank for the census scheduled for this year. As a rule, the greatest difficulty in securing this form of financing was at the level of national governments. The directors of statistics have had to make a major effort to ensure that decision-makers responsible for loans are fully aware of, and understand, the importance of the census. This is no easy task, bearing in mind that even if loans are secured, the cost of the census is still a fiscal expenditure, not to mention the fact that on occasions it has to compete with extremely important social projects at the national level.

27. Another type of innovative effort which has been observed recently is the possibility of countries’ sharing the cost of a certain instrument necessary for the census or making it accessible to the other country concerned. This has been applied, for example, to the use of optical mark readers or scanners, which may be shared by two or more countries. This idea also has a more important dimension, which is the work of census coordination among countries. One case in point is the above-mentioned census coordination programme between the Mercosur countries.

28. In addition, one frequent concern is that of establishing a policy of selling information, which implies cost recovery, mainly when the data are required by private firms. In this regard, there may be a wide range of products (publications, CDs, special processes), which, depending on the requirements and the type of user, would be handed over free of charge or at differential prices. This source of funding may not be very significant, or would be obtained after the census has been completed, and, in many cases through internal provisions, the funds collected under this heading would go to a central fund and not necessarily to the statistical institutes.

4. Conceptual innovations

29. The discussion of the contents of censuses today is not unrelated to the challenges raised by society today or, as a result, to the demand for information that this implies. Current development strategies, which promote economic growth with social equity, are closely linked to the living conditions of the population, whether the latter is considered as a factor of change which requires an increase in productivity and adjustment to technological change, or as the target for economic and social progress. This has brought new demands through increasing requirements in terms of information, from both the public and the private sector (NGOs, universities, research centres and the production sector). The result of this is the involvement in the censuses of some new issues relating to the housing environment, the provision of household equipment and the living conditions of persons. Nevertheless, it should be noted that these changes have been more limited, reflected at times in a few additional questions. It should be emphasized that, rather than a thorough modification of the issues involved, there is much more intensive use of the information traditionally collected in the census exercise.

30. Questions on housing and the household are seen not just as indicators of living conditions from the economic viewpoint and ways of estimating the housing deficit, but also as part of the environmental situation of the population. In general, the traditional questions give a general idea of sanitary conditions: access to drinking water, availability of sewerage systems, the number of bedrooms (overcrowding), and fuel used for cooking purposes. Since the 1990s, some countries have asked questions on the method of household waste disposal, and, of the four Latin American countries which conducted a census in 2000, three included this question in this round. In the preliminary discussions, other subjects were raised, including the type of road surface, street sweeping and the existence of green spaces in the vicinity, but no final decision was taken to incorporate them in the questionnaires used for census exercises that have already taken place.

31. With respect to provision of household equipment, many of the appliances considered earlier (refrigerator, television set, etc.) are no longer considered indicative of any clear differentiation in terms of social stratification. Hence, countries now include those appliances that suggest that the household has kept abreast of development and technological progress and all that this implies. The two most common items included in the censuses are the availability of a mobile phone and microcomputers and, in the case of the latter, there are countries which plan to investigate access to the Internet.

32. As regards new concerns relating to the questionnaire on individuals, increasingly, questions are included on health and social security coverage. Since reforms being carried out in Latin America are geared towards privatization, these questions reflect interest in finding out the extent of coverage provided by public systems and private health-care institutions and pension fund administrators.

33. Under the questions relating to education, in addition to the traditional questions on literacy, attendance at school and the level of education or last full year of schooling attained, in the censuses of recent decades, other interesting aspects were incorporated. On the one hand, the investigation of pre-school education is becoming usual, since this level of education has been spreading in most countries. On the other hand, several countries have included a question on the professional title and postgraduate specialization. It should also be noted that some countries have incorporated informal vocational or technical education. These last issues are related to workforce studies based on skills, although the results obtained might not be absolutely reliable.

34. On the whole, questions on the economic characteristics of the population are broader, normally covering activity status, occupation, occupational category and branch of activity. An important number of countries have incorporated working hours and others the size of the establishment, the workplace and/or family income. Census definitions are not always consistent with those used in continuous household surveys, but there is consensus that the latter are a better instrument for studying in depth certain aspects of economic activity. With respect to the activity status, the newest proposals for censuses are generally aimed at more accurate data capture on the active population, especially for those who are involved in informal jobs, which are often those held by women. In the 1990s, eight countries included a further question on this point and three investigated the existence of a cottage industry in the questions addressed to the household. Since the 1991 census, Argentina has presented the options in separate questions, placing emphasis on activity versus inactivity (Giusti and Rodriguez Gauna, 1999). In the same vein, for the next census, a question design is being proposed to improve the quality of responses on status in employment. Once again, the proposal is to include, instead of a single multiple-choice question, a range of simple questions for each dimension implicit in the definition of occupational category. The authors cited highlight five dimensions that may be taken into account: the type of “economic risk”, the “type of authority”, the “origin of income”, the “responsibility of fiscal charges” and “the type of relationship with the owner or employer”.

35. There are not many new questions on fertility and mortality. It has been proposed that research on maternal mortality be undertaken by including a question on maternal deaths in the household section. In particular, Bolivia is planning to include this question in the census form.

36. Another important effort concerns the most appropriate form of investigation on the ethnic origin of persons, an issue which was already included in earlier censuses. In the last census taken prior to 2000, 13 countries sought to identify the indigenous population using different criteria (Peyser and Chackiel, 1999). The importance of investigating this issue through the census has been widely recognized and accepted, both because of the universality of the census as a source of data and because of the opportunity it affords to establish a relationship between ethnic origin and other characteristics of persons and households. The interest in this issue is a reflection of the growing demand coming from the ethnic groups or communities themselves, who view it as a means of increasing their social profile and obtaining greater recognition and resources; it may also enable them to obtain information for their own activities as a group and for government agencies or non-governmental organizations in the preparation of specific social policies designed to improve the quality of life of these communities.

37. In recent meetings where the issue was discussed and, in particular, at the seminar “Todos contamos”, organized by the National Administrative Department of Statistics (DANE) and other agencies in Cartagena de Indias, Colombia, in 2000, special attention was given principally to the following: (1) applying participatory methodologies for the preparation of census questions in which the communities themselves intervene and obtain an appropriate conceptual definition of the identity of ethnic groups, which may be applicable both to indigenous populations and to groups of African origin, which are quite significant in some countries of the region; (2) discussing operational aspects of the census investigation and evaluating the importance of the different available alternatives (a single question, a series of questions, an alternative census on ethnic groups, qualitative studies, etc.); and (3) assessing the size of ethnic groups and establishing a sampling frame for in-depth studies on specific issues that require special attention, such as living conditions of indigenous communities and other ethnic minorities. The four countries that have held their censuses have studied indigenous population in some way.

5. Technological and operating innovations

38. The list of technological innovations, relating mostly to computers, would be far too lengthy and complex for the author and perhaps even for the reader. Thus, reference will be made only to the most significant points of the different phases of the census: preparatory phase, enumeration and post-census phase.

39. In the preparatory phase, an innovation is the use of digital mapping for censuses (that is, maps stored on electronic media), accompanied in some cases by the use of satellite imaging. Specialists state that the fundamental change is to stop using paper maps and start to work on the computer. Some of the potential advantages of digital mapping are: (1) permanent updating of files; (2) reproduction on any scale of a file using the same original; (3) the construction of a geographic information system linking mapping with database registers; and (4) computer-assisted segmentation for assigning work to census takers (De Gregorio, 1998). This innovation, applied in many of the countries of the region, also facilitates the work of enumeration and the possibility of disseminating results through thematic maps.

40. Until now, there was a general conviction that the most advantageous method for census training was the cascade method, which implied a chain transmission of training from the highest level down to the enumerators. From more recent discussions, it emerges that this method is less suitable than previously thought, in view of the distortions to which concepts may be exposed at each phase of transmission. Countries are therefore designing procedures for more direct transmission of concepts and methods for filling out questionnaires, aided by the most modern audio-visual methods. It has also been considered that the use of fewer enumerators, implying that the census would take several days, could allow for better training.

41. Since its 1993 census, Colombia introduced a form of census training through the country’s educational system (Arenas, 1998), which is now being adopted in varying degrees by other countries. Curriculum training, as it is termed, consists of establishing an agreement with the Ministry of Education whereby pupils in the last grades of secondary school are given a special course a few hours each week, during the six months preceding the census, to impart knowledge in statistical education and on the population census. This training includes activities to familiarize the student with census requirements. Surveys will be planned and conducted in the neighbourhood and then the census questionnaire can be handled more smoothly and more advisedly. The teachers in the formal school system are trained for this task by training officers who, where possible, are from the same statistical office.

42. Brazil points to the system of indicators of census management for data collection as an innovation in the census carried out in 2000 (Brazilian Geographical and Statistical Institute (IBGE), 2000). Integration is identified as the most important characteristic, bearing in mind that it brings together all the regions of the country and problems can be dealt with and solved promptly while the fieldwork is being conducted. This was possible thanks to the availability of notebook computers—including palm-held computers—distributed in the census areas, and to use of the Internet. The area coordinators were thus able to communicate immediately any data-collection problems as well as the quantitative data on the progress of the work, such as the total number of residences and population, by sex, enumerated daily. This information was automatically evaluated. At the same time the system facilitated the authorization for payment of the enumerators. In previous censuses, the preliminary results took at least three months, whereas with this method, they were available immediately as soon as the work was completed. In short, the system made it possible to obtain better-quality information and in record time.

43. In recent decades, optical readers were distributed for reading marks and thus facilitating the work of information input on magnetic media from the questionnaires. However, there was still a significant amount of work by hand involved in the codification of open questions (place of birth, occupation, branch of activity) (Ellis, 1998). In Latin America, Uruguay made the major technological advance in its 1996 census of adopting for this work scanners which could interpret marks as well as alphanumeric characters and thus were capable of reading handwritten texts. This speeded up the work of data entry and facilitated codification and automatic evaluations. This caused a complete revolution in the way the data compilation work was organized, minimizing the need for intermediate staff and allowing more efficient quality control of all processes (Niedworok, 1999). On the basis of this experience, most of the countries are incorporating this technology in their censuses for the 2000 round. The use of scanners is, however, not problem-free and consideration should be given to the need for special paper, design and printing of the questionnaires and special care taken in filling them out in the field.

44. The greatest diversity exists at the stage of use and dissemination of information in terms of possible uses of information and methods of storage and distribution. The most noteworthy option is to set up a database for microdata for quick retrieval of the desired information and even the possibility of producing tables “online” through the intranet or Internet.

45. To promote effective use of data, CELADE has created one of the most widely used data storage programmes in Latin America and the Caribbean: retrieval of data for small areas by microcomputer (REDATAM)[1]. Currently, the fourth generation version of the software (REDATAM+G4) is available for use in English, Portuguese or Spanish with Microsoft Windows 95, 98, NT4 or 2000, in any IBM-compatible computer. The software uses a compressed database for microdata with the registers of individuals, sets of housing, city blocks or any administrative division identified in the census. On the same basis, one or several censuses, surveys and administrative registers may be combined. Any geographical area of interest or combinations of such areas may be defined through a database to create new variables and rapidly show tabulations in graphic windows. The data from different geographic levels may be combined by order of importance to create aggregate variables and the results may be displayed on maps from REDATAM or transferred to a geographic information system (GIS) (CELADE, 2000a, www.eclac.cl/celade/redatam/). In this way, population data and information of another type, such as available health centres or educational establishments, may be included on the same map for a particular geographical unit.

46. In addition, REDATAM contains modules for generating and administering databases and includes R+ ZonPlan , which creates predetermined indicators. Currently, another module is being established (R+ Xplan) to enable users to create their own indicator applications along the same lines as ZonPlan. On the strength of these experiences, CELADE and IDB recently designed a project for the dissemination or use of census data through the construction of this type of database.

6. Census evaluation

47. Census evaluation consists of analysing the quality of work at all stages of the census exercise, as well as the quality of the results obtained. In the following section, reference is made basically to the evaluation of the census coverage. On the whole, two procedures are proposed for evaluating census coverage: the direct method based on the post-enumeration survey and the indirect method based on the exercise referred to as census conciliation.

a. The post-enumeration survey

48. This procedure consists of conducting a sample survey which is used to evaluate the census coverage and, in some cases, also to check some of the data investigated in the census. This survey must have the following characteristics:

· The survey must be conducted within a few days after the census to avoid the problems of population mobility or other demographic changes; and

· In principle, it must be done by an entity independent from the one that conducted the census and by staff that did not work on the census enumeration, which is linked to the institutional independence and to the assumption of statistical independence that the methodology demands.

49. The most widely accepted evaluation method based on the survey consists of matching the census form and the survey questionnaire. Since it is assumed that the enumeration of one person in the survey is independent of what may occur in the census, and since both sources are subject to error, it is possible to identify those persons registered in both sources and those captured by one source and not the other. Then, through a statistical formula, it is possible to work out the probability of omission from both sources and thus to establish an estimated total as the accurate one (Marks, Seltzer and Krótki, 1974).

50. While these operations are carried out by most countries (14 countries in the censuses for the 1990s, Table 2), countries themselves, when surveyed and at the 1998 seminar, identified the following problems encountered with the procedure:

· Surveys are costly and distract attention from the main activity, which is the population census;

· In order to ensure that the surveys are valid, they must be done by an institution other than the one responsible for the census, and this applies even to geographic areas with independent mapping, which is not a common occurrence;

· Often the surveys are not conducted until long after the census, so that the results are no longer comparable;

· In some cases, the results of these surveys are dubious and there is no available methodology for calculating the rate of omission or for establishing a transparent technical report. This is serious because it means that the census loses its credibility; and

· Rarely are the results used exhaustively—for example, to derive information on the geographically disaggregated coverage or for analysing errors of content at least in some variables.

a. Census conciliation

51. The method of census conciliation for determining census coverage is not a routine mathematical statistical procedure but rather an analysis of the information with demographic criteria based on the knowledge and cumulative experience on problems relating to information. The final result is a coherent description of the population dynamic considered as the most plausible for a given period which includes the available censuses. This dynamic includes the estimation of the population by sex and age and the growth components that account for its variation (fertility, mortality and international migration). The result of this exercise is the change in population in a particular period of a country’s history (for example, 1950-2000), which is used as the basis for working out population projections by sex and age.

The procedure is based on two central ideas:

1. If the censuses were taken without errors and were accurate estimates of mortality, fertility and international migrations, then it would be possible, on the basis of the population by sex and age of one census, to calculate that corresponding to another earlier or later census, which would coincide with the values observed. For example, the population aged 20-24 years in the 1990 census may be calculated as the population aged 10-14 years in the 1980 census multiplied by the survival ratio of this group over the 10-year period, plus net international migration of the cohort during the period.

2. In reality, censuses are subject to errors of coverage and age-misreporting, which follow relatively well-known patterns. For example, experience in Latin America shows that there is a significant omission of the population of children aged 5 years or under and of working-age men, an overreporting of the age by elderly persons, an underreporting of the age of adult women, and so forth. This causes greater confidence in the population enumerated in certain age groups for each sex as well as greater doubts about the actual situation for other groups.

52. The methodology consists of calculating the different independent population estimates by sex and age groups which may be obtained for the date of the census which is to be evaluated. Where three censuses are available, for the particular census for which the evaluation is required, for each of the sexes, six estimates would be available: those relating to the same sex from each one of the censuses and those relating to the opposite sex. Then it would be necessary to select the value or average of values considered most reasonable.

53. The purpose of the procedure described here is not to obtain the absolute truth on the population dynamics of a country. The objective is to obtain the most plausible estimates with the available data, ensuring at least the existence of consistency between the variables involved. Undoubtedly, consistency is a prerequisite, although not sufficient in itself to yield accurate estimates. Perhaps the greatest drawback is that it depends to a great extent on the good judgement of the evaluator and in this regard there is a high degree of subjectivity. The greatest difficulties are linked to problems of lack of comparability of the censuses and above all to the fact that it is difficult to determine the net migratory balance between censuses.

54. In conclusion, since most of the countries conduct post-enumeration surveys, the two procedures are considered to be complementary. The surveys are treated as forming a part of the information available for the evaluation, but it would be necessary in any event to obtain a plausible demographic explanation of the census results.

B. Conclusion

55. Censuses have an important role to play in Latin America as they are the basis for the national statistical system and are the main source of information for small subpopulations. While administrative registers have improved and there is supposedly a technological basis for proper administration of their data, they are still far behind the population registers maintained in the developed countries.

56. The high costs involved and the shortage of public funding threaten the very survival of the census. Alternatives to the traditional sources of funding must be found to ensure that censuses are held and the desired standards are met. Thus, it is recommended that attempts be made to involve the private sector and to harness international funding through loans, for example, from the IDB, which means that the country would assume the cost but pay it over several years. Another alternative presented is that of sharing certain costs between two or more countries, for example, purchasing equipment and realizing economies of scale.

57. The 2000 round of censuses is characterized by major changes, principally at the technological level. From the conceptual point of view, investigations are continuing on practically the same concepts as in the past, although probably some questions are being incorporated relating to current concerns that affect the population. From the technological viewpoint, on the other hand, reference may be made to various new factors that may affect the timeliness and quality of data: digital mapping, data capture using scanners and use of the Internet to manage the census in the field and to disseminate census information.

58. Lastly, it is crucial that censuses should yield high-quality results. The availability of highly advanced technological tools makes it possible to do so without the risk otherwise associated with greater rapidity in obtaining results. In order to evaluate this point, the census coverage should be studied in conjunction with direct as well as indirect procedures. The direct results are obtained through the post-enumeration survey and the indirect method seeks to establish consistency with the information derived from the population dynamics within the country.

References

Arenas, R. (1998). Capacitación censal, Cartagena de Indias, National Administrative Department of Statistics (DANE)/ United Nations Population Fund (UNFPA).

Chackiel, J. (1999). Mortality in Latin America. Health and Mortality: Issues of Global Concern (ST/ESA/SER.R/152), New York.

De Gregorio, H. (1998). Cartografía digital para planificar relevamientos estadísticos. Aquí se cuenta. Revista informativa del censo 2000, No. 2, Buenos Aires, National Institute of Statistics and Censuses (INDEC).

Del Popolo, F. (1999). Encuesta sobre experiencia censal en aspectos conceptuales. América Latina: aspectos conceptuales de los censos del 2000, Manuales series, No. 1, (LC/L.1204-P), Santiago, Chile. United Nations publication, Sales No. S.99.II.G.9.

ECLAC (Economic Commission for Latin America and the Caribbean) (1999). América Latina: aspectos conceptuales de los censos del 2000, Manuales series, No. 1, (LC/L.1204-P), Santiago, Chile. United Nations publication, Sales No. S.99.II.G.9.

ECLAC-CELADE (Economic Commission for Latin America and the Caribbean, Population Division, Latin American and Caribbean Demographic Centre (2000a). Redatam Informa” (LC/L.1466), Santiago, Chile.

(2000b). International Migration in Latin America, IMILA. Demographic Bulletin, year 33, No. 65 (LC/G.2065-P), Santiago, Chile.

Ellis, C. (1998). Nuevas Tecnologías en captura y diseminación de datos de población. O censo 2000 no MERCOSUL. Segundo Seminário de Trabalho, Rio de Janeiro, Brazilian Geographical and Statistical Institute (IBGE).

Giusti, A. and M. C. Rodriguez Gauna (1999). Dos desafíos de los censos argentinos: la medición de la condición de actividad en el censo de 1991 y la categoría ocupacional en el censo del 2000. América Latina: aspectos conceptuales de los censos del 2000, Manuales series, No. 1, (LC/L.1204-P), Santiago, Chile. United Nations publication, Sales No. S.99.II.G.9.

IBGE (Instituto Brasileiro de Geografia e Estatística) (2000). Vou te contar. A revista do censo 2000, No. 3, Brasilia.

Marks, E. S., W. Seltzer and K.J. Krótki (1974). Population Growth Estimation. New York: The Population Council.

Niedworok, N. (1999). Algunas reflexiones acerca de un formulario censal diseñado para captura de datos mediante tecnología de imágenes (scanners) en el censo del 2000. América Latina: aspectos conceptuales de los censos del 2000, Manuales series, No. 1, (LC/L.1204-P), Santiago, Chile. United Nations publication, Sales No. S.99.II.G.9.

Peyser, A. and J. Chackiel (1999). La identificación de poblaciones indígenas en los Censos de América Latina. América Latina: aspectos conceptuales de los censos del 2000, Manuales series, No. 1, (LC/L.1204-P), Santiago, Chile. United Nations publication, Sales No. S.99.II.G.9.

Schkolnik, S. and Guzmán, J.M. (1999). América Latina: los censos del 2000 y el desarrollo social. América Latina: aspectos conceptuales de los censos del 2000, Manuales series, No. 1, (LC/L.1204-P), Santiago, Chile. United Nations publication, Sales No. S.99.II.G.9.

* This document was reproduced without formal editing

** ECLAC, Chile. 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

[1] This software is not intended to compete with, but rather to complement, the Integrated Microcomputer Processing System (IMPS) of the United States Bureau of the Census.

Symposium 2001/14

Juan Chackiel **