PFII/2004/WS.1/13 Original: English

UNITED NATIONS NATIONS UNIES

DEPARTMENT OF ECONOMIC AND SOCIAL AFFAIRS Division for Social Policy and Development Secretariat of the Permanent Forum on Indigenous Issues

WORKSHOP ON DATA COLLECTION AND DISAGGREGATION FOR INDIGENOUS PEOPLES (New York, 19-21 January 2004)

Data collection pertaining to indigenous peoples: issues and challenges By Jeremiah P. Bandah1

Introduction

1. The Permanent Forum on Indigenous Issues in its report on the first session called upon all concerned agencies and organisations to begin to disaggregate data on indigenous people generally, and indigenous women and children specifically, in two categories covering (i) programmes and services impacting indigenous peoples, and (ii) fiscal allocations for indigenous peoples’ programmes and services (United Nations, 2002, para.6).

2. Some basic data on indigenous people are available in a number of countries. Several countries, for example, use population and housing censuses to gather information on selected demographic and socio-economic characteristics of indigenous people. Obtaining relevant and comparable statistical information on indigenous people, however, still remains a challenge in many countries.

3. The aim of this paper is to provide an overview of some of the main issues and challenges related to the collection of data on indigenous people. Particular emphasis is given to the three major sources used to collect these data and to some of their main advantages and drawbacks. The paper also provides an overview of the data on national and ethnic groups collected and disseminated through the Demographic Yearbook system. National data collection practices specific to, among others, ethnicity, ancestry, aboriginal and indigenous groups in the 2000 round of censuses are also presented. II. Data sources

4. There are three main sources of statistics that are used to collect information on indigenous peoples, namely, censuses, sample surveys and administrative records. We briefly discuss these sources in the following paragraphs.

2.1. Population and housing censuses

5. A population census, hereafter simply referred to as census, is the total process of collecting, compiling, evaluating and disseminating demographic, social and other data at a specified time covering all persons in a country or in a well-delimited part(s) of a country (United Nations, 1998). A census is part of the fundamental statistical base of a country. It can, therefore, be a major source of social statistics, with its obvious advantage of providing data for small geographic units. A census is an ideal method for providing information on size, composition and spatial distribution of the population in addition to socio-economic and demographic characteristics. In general the census collects information for each individual in households and each set of living quarters, usually for the whole country or well defined parts of the country. The basic features of a population and housing census include:

i. Individuals in the population and each set of living quarters are enumerated separately in the population.

ii. Universality within a defined geographic area/territory. The population census potentially covers the whole population in a clearly defined territory. It should include every person present and/or usual residents depending on whether the type of population count is de facto or de jure. In the absence of a comprehensive population and administrative register it is the only source that can provide small area statistics.

iii. The enumeration has to be as simultaneous as possible. The persons and dwellings should be enumerated with respect to the same well-defined reference period.

iv. Censuses are usually conducted at defined intervals. Most countries conduct censuses after every 10 years while others after five years. This facilitates the availability of comparable information at fixed intervals.

v. Censuses provide information on size, composition and spatial distribution of the population together with demographic and social characteristics. Such information is usually available for small administrative domains/ areas.

vi. They are also sources of sampling frames.

6. While censuses have the advantage of providing information for smaller geographic areas they cannot be repeated frequently, as a result information collected through censuses can become outdated. It is against this background that, household surveys provide a basis for updating some census information especially at national and other large domain levels. In most cases only relatively simple topics are investigated in a census and the number of questions is usually limited. Census information can therefore be complemented by detailed information on complex topics from the household surveys taking advantage of their small size and potential flexibility.

2.2. Sample surveys

7. Sample surveys are a key source of data on socio-economic phenomena. They are among the most flexible methods of data collection. In theory almost any subject can be investigated through sample surveys. In this respect, specialized information pertaining to indigenous issues can ably be collected through surveys. In sample surveys part of the population is selected from which observations are made or data are collected and then inferences are made to the whole population. With the potential of smaller workloads than in censuses and the possibility of training fewer people more intensively, household surveys can cover most subject matters in greater detail than censuses. As it is not possible to meet and anticipate all data needs of a country through a census, household surveys provide a mechanism for meeting the additional and emerging needs on a continuous basis. The flexibility of household surveys, therefore, make them excellent choices for meeting data users’ needs for statistical information which otherwise would not be available, insufficient and perhaps unreliable.

8. There are different types of household surveys that can be conducted to collect data relevant to indigenous concerns, such as specialised surveys, multi-phase surveys, multi-subject surveys and longitudinal surveys. The selection of a specific type of survey will depend on a number of factors including, subject matter requirements, resources and logistical considerations. Specialised surveys cover single subjects or issues such as time- use, disability, nutrition etc.

2.3. Administrative Records

9. Different types of statistics can be compiled from various administrative records as by-products of administrative processes. Examples include health statistics compiled from hospital records, employment statistics from employment exchange services, vital statistics compiled from the civil registration system and education statistics from enrolment reports of the ministry of education.

10. The reliability of statistics from administrative records depends on the completeness of the administrative records and the consistency of definitions and concepts. It is therefore necessary to continuously improve and update the systems of recording, compiling and analysing of such data. Wherever possible, it is advisable to use the same definitions and concepts overtime and harmonize these with the definitions and concepts used in other data sources. This would facilitate the comparability of data. 11. While administrative records can be very cost-effective sources of data, such systems are not well developed in many countries. This implies that in a majority of cases such data are unreliable. Even if the administrative recording processes are continuous for purposes of administration, the compilation of statistics is, in most cases, secondary. Statistical requirements that need to be maintained such as standardisation of concepts and definitions, adhering to timeliness and complete coverage are not usually considered or adhered to.

12. For most countries, information from administrative records is often limited in content as their uses are more specifically designed for legal or administrative purposes. A civil registration system is an example of an administrative system that many countries have developed but with varying success. Countries with complete vital registration systems, which include a component of vital registration, are able to produce periodic reports on vital events, such as number of live births by sex, date and place of births, number of deaths by age, sex, place of deaths and cause of death, marriages and divorces etc.

13. A population register maintains life databases of every person and household in a country. The register is updated on a continuous basis when there are changes in the characteristics of individuals and households. If such registers are combined with other social registers they can be a source of rich information. Countries, which have developed such systems, include Denmark, Norway, the Netherlands, Germany and Sweden. For most of these countries censuses are based on the registration system.

14. In many developing countries, while administrative records for various social programmes can be cost-effective data source and an attractive proposition, they are not well developed. In this case their complementary use with other sources is a big challenge because of lack of standardized concepts, classification systems coupled with selective and under coverage.

III. Data collection and dissemination through the Demographic Yearbook system

15. The United Nations Statistics Division collects and disseminates data, mainly demographic data, and in some cases national and ethnic composition of populations. Although most of these data do not always explicitly identify indigenous peoples, they provide some indication of such groups.

16. Data on ethnic groups has been included in the 1956, 1963, 1964, 1971, 1979, 1983, 1988 and 1993 editions of the Demographic Yearbook. These statistics are based on results from population and housing censuses.

3.1. Diverging terminology

17. The 1956 and 1963 editions of the Demographic Yearbooks examined indigenous populations indirectly in the context of “ethnic composition”. The 1956 Demographic Yearbook noted that these data “fail to take account of the many ethnic- subgroups with a diversity of culture patterns. They fail also to reflect the lack of assimilation in some of the native born who maintain the cultural habits of their forefathers”. Another common problem associated with these data is the lack of international comparability.

18. Since 1979 these data have been labeled as “national/or ethnic group”. The technical notes included in the Demographic Yearbook, underscore the heterogeneous nature of the statistics related to ethnicity by defining them as a single or combined category of race, religion, colour, stock, tribe, ethnic origin, or ethnic nationality.

19. Recognizing the multiple interpretation of the concept of ethnicity, the 1998 Principles and Recommendations for Population and Housing Censuses lists a wide range of criteria that are or can be used to identify ethnic groups. These include ethnic group and nationality, race, colour, language, religion, customs of dress or eating, tribe or various combinations of these characteristics. These criteria can have different meanings for different countries and/or at different points in time. It is therefore, imperative; that countries clearly specify the criteria used in the census reports such that the meaning of the classification is discernible.

3.2. Uses of ethnicity data

20. Despite the limitations of international comparability of data on ethnicity, due to the use of different concepts in different countries that report these statistics there are some general usage of such data, for example:

i. Data on population subgroups are essential in analysing national socio- demographic characteristics; and

ii. Statistics on ethnic origin constitute a framework for demographic analysis and social studies.

21. The United States Government for instance has collected data on race and ethnicity since its first census of population in 1790 (Evinger, 1995). The data have been used to study changes in social demographic, health, and economic characteristics of various groups of people. The data from censuses, surveys and administrative records have provided a historical record of the population diversity and its changing attitudes and policy concerns about the identified population groups. 3.3. Limitations of ethnic data

22. The major limitation in comparing data on nation and/or ethnic groups, at international level, is the variability between different countries in collection of data on the subject. Even in relation to the usage of the term “race” the definition and amount of detail used to identify race may differ significantly among countries.

23. Another constraint is that in most cases the data are unreliable. For example, self- identification with a certain ethnic group may be clear and well defined in some cases, such as indigenous groups that have existed for many generations. Alternatively in other cases the respondent may feel that he/she does not fit in one group or might identify with more than one ethnic group, all of which may lead to confusion. Measurement error is another source of unreliability of data on ethnicity. In censuses it matters whether:

i. The response is provided by a family member or is inferred by the interviewer.

ii. Legal rights or acquired benefits are perceived. This may cause, in some cases, response bias in studies.

iii. In some countries changes occur in ethnic identification.

iv. There is undercoverage in the censuses.

24. It has been reported for instance, in the Untied States of America that people identify with more than one race or ethnic group. Evinger (1996) noted that, for example, people with one Hispanic parent and one non-Hispanic parent may say yes to a separate Hispanic-origin question but may not state that Hispanic in their sole identification.

3.4. Coverage of ethnicity data in the Demographic Yearbook system

25. Table 1 uses the information obtained from the Demographic Yearbooks to show the number of countries that reported on “national and/or ethnic groups” for each census round and by continent. Countries that held more than one census in a given round have been counted only once. Due to changes in geographical boundaries, these numbers cannot be compared from one census round to another.

26. In all of the above-mentioned issues of the Demographic Yearbook, most countries reported the statistics on “National and/or ethnic group” by sex. However, in each census round there was a small group of countries for which a breakdown by sex was not available.

27. In addition, starting from the 1971 Demographic Yearbook, a new classification with an urban/rural variable was included. The following are the countries or areas that reported on “National and/or ethnic group” by sex and by urban/rural areas: Guatemala, South Africa and Southern Rhodesia (in the 1971 Demographic Yearbook); United States, Australia, and Macao (in the 1973 Demographic Yearbook); Sri Lanka (in the 1979 Demographic Yearbook); and Belize and Romania (in the 1993 Demographic Yearbook). In 1983 and 1988, there were no countries that included an urban/rural classification.

Table 1: Coverage of ‘National and/or ethnic group” in the Demographic Yearbooks

America, America, Africa Asia Europe Oceania Total Census round DYB North South (N) (N) (N) (N) (N) (N) (N) 1950 1956 23 18 3 13 2 15 74 (1945-1954) 1960 1971, 1964, 44 20 2 10 6 18 100 (1955-1964) 1963 1988, 1970 1979, 17 12 1 10 7 17 64 1973, (1965-1974) 1971 1980 1988, 1983, 10 19 2 10 9 16 66 (1975-1984) 1979 1990 1993, 6 5 1 7 5 5 29 1988 (1985-1994)

(N): number of countries that reported information on “National and/or ethnic groups”

IV. National data collection practices in the 2000 round of censuses

28. In order to explore in more detail some of the conceptual and methodological issues surrounding the concept of ethnicity its implicit reference indigenous peoples, we refer to a review of national data collection practices from the 2000 round of censuses (Alemenay and Zewoldi, 2003). A classification – based on the different question formats and concepts used in the censuses – was derived from the information provided in the census questionnaires.

4.1. Countries and Areas included in the study

29. The 2000 round of censuses covers the period from 1995 to 2004. The findings reported in this paper are based on the analysis of census questionnaires from 147 countries or areas, which represent 79 percent of the countries that have already conducted a census during the 2000 round. From the 228 countries or areas recognized by the United Nations, 81 were excluded from this analysis. The reasons for their exclusion are as follows: twenty three have not taken or planned a census for the 2000 round (nine of them have population registers instead), nineteen are expected to have a census in the future, and 39 have supposedly completed a census but the questionnaires were not available at the time of compiling the figures. Table 2 below shows the number of countries or areas that were included and those that were excluded by continent.

30. The coverage rates, defined here as the percentage of countries that were included in the analysis from the total number of countries that completed a census (that is excluding the category of non-available countries), are the following: 17 out of 39 countries (44 per cent) in Africa; 30 out of 34 (88 per cent) in North America; 11 out of 13 (85 per cent) in South America; 34 out of 39 (87 per cent) in Asia; 36 out of 37 (97 per cent) in Europe; and 19 out of 24 (79 per cent) in Oceania. Overall, this study comprises 79 per cent of the countries that have conducted a census in the 2000 round.

Table 2: Countries And Areas covered

America, America, Africa Asia Europe Oceania Total North South N % N % N % N % N % N % N % Questionnaires included 17 30.5% 30 81% 11 79% 34 71% 36 75% 19 76% 147 64% Missing Questionnaires* 22 39% 4 11% 2 14% 5 10% 1 2% 5 20% 39 17% Non-available Questionnaires** 17 30.5% 3 8% 1 7% 9 19% 11 23% 1 4% 42 18%

TOTAL 56 100% 37 100% 14 100% 48 100% 48 100% 25 100% 228 100%

* “Missing questionnaires” refer to those countries that completed a census but for which questionnaires were not available

** “Non-available questionnaires” refer to those countries in which no census has been taken or planned or in which a census is expected to be taken at a future date, and therefore a questionnaire was not available.

4.2. Coverage of ethnicity in the 2000 round of censuses

31. From the 147 countries or areas included in our analysis, 95 countries (65 per cent) asked one or more questions on ‘Ethnicity’. Table 3 shows the distribution of countries by continent.

Table 3: Coverage in the 2000 round of censuses of "national and/or ethnic group"

Number of Number of Countries that did Countries that countries countries NOT include included 'National included in that 'National and/or and/or ethnic group' the conducted a ethnic group' analysis census N % N % N N Africa 7 41% 10 59% 17 39 America, North 23 77% 7 23% 30 34 America, South 9 82% 2 18% 11 13 Asia 22 65% 12 35% 34 39 Europe 19 53% 17 47% 36 37 Oceania 15 79% 4 21% 19 24 Total 95 65% 52 35% 147 186

32. As shown in table 3, South America is the continent with a higher proportion of countries including a question on ethnicity, closely followed by Oceania and North America. Asia and Europe had lower proportions, although still more than half of the countries included an ethnicity question. Finally, 7 out of the 17 African countries asked about ethnic groups in their censuses. Please note again that due to the low number of African countries included in the study (44 per cent of those that conducted a census), these percentages may not give an accurate idea of the reality and therefore cannot be generalized to the whole continent.

4.3. Concepts and terminology used in censuses

33. As previously mentioned, the criteria for defining ethnicity usually differ significantly from one country to another and from one point in time to another and may be based on factors such as: colour, race, ethnic nationality, language, religion, customs, tribes, etc. This heterogeneity is clearly reflected in the terminology used in the 2000 round of censuses.

34. For classification purposes, all questions used in the 2000 round of censuses were combined into 6 different categories: “ethnic group”, “ancestry or ethnic origin”; “race”; “nationality”; “indigenous or aboriginal groups”; and “tribes” and/or “castes”. However, the delineation between categories is not always well defined and in some cases census’ questionnaires included combinations of two or more of these groupings. The classification shown in table 4 is based on terminology used in the questions and not the concepts in which the output categories are based. However, comments on the discrepancies between the terminology and the underlying concepts are included in this section. Table 4 shows the detailed distribution of countries by terminology and continent.

Table 4: Terminology used in the census questionnaires by number of countries

America, America, Africa Asia Europe Oceania Total North South N % N % N % N % N % N % N % -- Question on 6 86% 16 70% 3 - 12 55% 7 37% 7 47% 51 54% 'Ethnic group' -- Questions on 'Ancestry' or ethnic - - 3 - - - 1 - - - 3 - 7 7% origin' -- Question on 'Race' - - 6 26% 1 11% - - - - 4 27% 11 12% -- Question on - - 1 - - - 7 32% 12 54% 1 7% 21 22% 'Nationality' -- Question on 'Indigenous/aboriginal - - 4 - 6 67% - - - - 2 - 12 12% groups' -- Question on 1 - - - - - 2 - - - 1 - 4 - 'Tribes' and/or 'Castes'

%: Column percentages are calculated within a given continent. In some cases, they do not add up to a 100% because some countries included more than one question on ethnicity.

Ethnic group

35. As table 4 shows, slightly more than half of the countries for which a questionnaire was available included a question with the term “ethnic group”. In most cases, the question format was “To what ethnic group does (the person) belong to?” Although the terminology used in the question format was apparently the same for all these countries or areas, the response categories referred to a variety of concepts such as race, nationality, indigenous groups or a combination of two or more of those. In Africa, 3 out of the 6 countries that asked about “ethnic group” included answer categories that referred to race2. In the case of Zambia and Senegal, “locals” had to enter their ethnic group and foreigners had to mark a major racial group (in Zambia) or their nationality (in Senegal). In North America, Costa Rica and Trinidad and Tobago included a question on “ethnic group” with response categories that reflected a racial classification. The same was true for the United Kingdom. Jersey (Channel Islands) and Guyana combined race and nationality in their response categories.

36. Both in North and South America, the indigenous component was included for some countries in their response categories. Belize, Peru and Suriname are three countries that used a combination of racial and indigenous categories in their response boxes to the “ethnic group” question. Honduras and Guatemala used the term “ethnic group” to ask for affiliation to indigenous groups. Although these countries were grouped under the “ethnic group” category, they will be discussed together with the rest of countries that included a question on indigenous or aboriginal groups.

37. On the other hand, countries that integrated two or more terms (one of them being “ethnic group”) in their question format were also grouped under this category. Examples are Antigua and Barbuda, British Virgin Islands, Montserrat, Saint Kitts and Nevis, and Saint Lucia, all of which asked, “To what ethnic, racial or national group do you think you belong to?” and combined the three concepts in their output categories.

38. Finally, Malaysia and Singapore are also two special cases in which language was the principal criteria used in determining the respondent’s ethnic group. This linkage between language and ethnicity becomes very clear in the way the question is asked: “To what ethnic/dialect group does (the person) belong to?”

Ancestor and/or ethnic origin

39. “Ancestry” is another concept used to measure the ethnic composition of a population. As Aspinall (2001: 831) argues, “ancestry or ethnic origin” somewhat differs from “ethnic group” in the sense that it is an “externally allocated” concept of identity which “focuses the question back in time and conveys an historical and frequently geographic context”, whereas “ethnic group” is a “self-perceived conception of social group membership”. He then suggests that “ancestry” is a more stable concept that produces lower gross rates whereas self-identified “ethnic group” often yields to higher confusion and more inconsistent reporting.

40. It is precisely to minimize confusion and ensure high quality data, that the Australian Bureau of Statistics included a question using the concept of “Ancestry”. In it was argued that “ancestry” had been chosen over other measures because the goal of such question was to identify the respondents’ origin rather than a subjective perception of their ethnic background (Edwards, 2003). In the 2000 round of censuses, Canada, Kiribati, Niue, Puerto Rico, and the United States were the other countries that asked a question using the term “ancestry” and/or “ethnic origin”. The Special Administrative Region of Macao, which asked a question on the ethnicity of the respondent’s parents, was also included under the “ethnic origin” category.

Race 41. As discussed in the section on ethnic groups, at least 18 countries included a question on “ethnic group” with answer categories referring to race or a combination between race and other concepts. In addition, 11 out of the 95 countries that included a question on ethnicity in their census used the term “race” in the formulation of the question itself. These countries were: Anguilla, Canada, Jamaica, Puerto Rico, United States and the US Virgin Islands (in North America); Brazil (in South America); and, American Samoa, Guam, Federated States of Micronesia, and Northern Mariana Islands (in Oceania). These last four countries included a question and response categories combining the concepts of “race” and “ethnic origin”.

42. The use of a racial classification in censuses to identify ethnic groups is a controversial issue currently being debated in some countries. It has been argued that racial categories carry strong connotations from colonial times and do not take into account the heterogeneity within each group (Aspinall, 2001). However, it also has important advantages when identifying those groups that have been historically excluded and marginalized, groups that may be less easy to identify through self-reported open- ended ethnicity categories (Aspinall, 2001).

Nationality

43. Another major concept used in the 2000 round of censuses to identify the ethnic composition of a population was ethnic “nationality” (different from legal nationality or country of citizenship). As table 4 indicates, 19 of the 21 countries that asked a question on ethnic “nationality” are to be found in Asia and Europe. In Europe, with the exception of Malta, the countries that asked a question on “nationality” were Eastern European countries, and 5 of them are former Soviet Union Republics. In Asia, the 7 countries that included a question on ethnic “nationality” were all from the Commonwealth of Independent States. Nauru (in Oceania) and Aruba (in North America) also ask a question on “nationality”. In Africa, although grouped under another category, Zambia and Kenya also asked a question on the foreigners’ nationality.

Aboriginal or Indigenous Groups

44. There has been a growing interest to include a question on indigenous groups in population censuses. Not only is the information relevant for economic, social and health policies, but it has also been used as a tool for indigenous communities to become more visible and to reinforce their identities.

45. Our classification identifies countries that have a specific separate question on indigenous identities. Therefore, it does not include under the “indigenous groups” category those countries (i.e. Puerto Rico, Saint Lucia, United States) that ask general questions on ethnicity or race and include an output category for indigenous people. As table 4 shows, the inclusion of a specific indigenous question is common mainly in Latin American countries and in Oceania. In North America Canada, Honduras, Guatemala, Mexico, and Panama included a specific question on indigenous groups. In South America Argentina, Bolivia, Chile, Paraguay, Suriname and Venezuela also included a question in their censuses. Peru, which is not included in the table as belonging to this group, had a question on “ethnic group” with output responses that mixed racial and indigenous categories. Finally, in Oceania, New Zealand had a question on Maori descendants and Australia asked a question on Aboriginal and Torres Strait Islander origin.

Tribes and/or castes

46. The knowledge of tribal characteristics, as noted in the 1983’s Demographic Yearbook technical notes, is “essential to any study of economic and social development in societies where tribal population is important”. The classification in table 4 does not reflect the measurement of such characteristics. Most African census questionnaires that may have asked about tribal characteristics (i.e. Ghana, Senegal, and Zambia) had open- ended questions on “ethnic group”. Therefore, the lack of accessibility to the enumerator’s manual and to the list of response categories does not shed light to whether the responses were measuring tribal affiliation or some other concept such as race. In Africa, only Kenya included a specific question with the term “tribe” in it. Nauru (in Oceania) and India (in Asia) also included a question on “tribes”. As for “castes”, India and Nepal were the only two countries that included a question in their censuses.

4.4. Question format

47. So far, we have explored the conceptual and terminology issues used in the measurement of ethnicity in censuses. This section will analyze the different formats of response categories and its methodological implications.

48. The level of detail and the typology of response categories differ from country to country. Racial classifications, for example, tend to imply fewer groupings (6 or 7 on average) than nationality3. Even when the concepts and terminology used is apparently the same; countries may utilize some groupings and categories that others may not. All this variation tends to limit international comparability of data from censuses.

49. The level of detail and consistency of responses also depends on the structure of the question itself. In this study, we have elaborated another classification that divides countries into five different groupings, based on the question format: open-ended question, question with response categories that do not include an open-ended “Other” box, question with response categories with the option of specifying what “Other” is, a Yes/No question, and finally an “unknown” type of answer for those countries for which the full questionnaire or the code book for possible answers was not available. Overall, 95 countries asked 107 questions on ethnicity. Forty-six out of the 107 questions (43 per cent) had pre-established response categories with an empty “Other” box (to be filled in by the respondent when appropriate). Twenty-one questions (20 per cent) had pre- established categories without an open-ended “Other” category. Twenty-three out of the 107 (21 per cent) were open-ended questions, four (4 per cent) were a yes/no question, and 13 questions (12 per cent) did not offer enough information on answer typology.

50. Another factor that may affect the accuracy of responses is the selection of answer categories or examples. An “example” effect takes place when the inclusion or exclusion of one of the example categories generates wide differences in the reporting. In 1993, the Canadian census included a “Canadian” category as an example in the ancestry question, resulting in an increase of self-reported Canadians from 763,000 to 5,193,000 (Aspinall, 2001). Therefore, the selection of response categories or examples has to be carefully selected in order to ensure an accurate reporting of ethnicity data. In addition, when the examples included in a question or the response categories change, data users should be careful when comparing data from one census to another.

4.5. Multiple ethnic identities

51. Migratory movements and intermarriage are two social phenomena that have increased the difficulty of measuring ethnicity. As a consequence of these demographic trends, multiple identities have increased, as well as the likelihood of reporting them, adding to the complexity of ethnic data (Waters, 2000). In the 2000 round of censuses, only 11 countries from the 95 that included a question on ethnicity allowed the option of selecting multiple ethnic identities.

52. In North America, Canada, the United States and the US Virgin Islands, for example, included the option of multiple entries. In the Canadian census respondents could mark as many identities as applicable both in the ancestry and race question. The United States and US Virgin Islands, which have a common question on race, also allowed for more than one categorisation. In addition, in the case of the United States an open-ended ancestry question provided the possibility of filling in one or two answers. In Oceania, seven countries allowed for multiple ethnic identities in the following questions: ethnic origin or race (in American Samoa, Guam, Marshall Islands, Micronesia, Northern Mariana Islands and Tokelau) and ancestry (in Australia). In Europe, only Hungary offered the option of selecting up to three ethnic nationalities. Responding to the demands of several minority representatives, the 2001 Hungarian census was the first to include the possibility of giving multiple answers on ethnicity.

53. As the demand for multiethnic identities become more common, there will be a growing need to adapt the census questions so that such phenomena is reflected in the results. For instance, the Russian Federation is facing increasing demands to include multiple ethnic identities in the 2012 census that would recognize ethnicities such as Russian Jews or Tatar-Russians that are currently not reflected in the census results (Tishkov, 2001).

V. Conclusions

54. The collection of data on indigenous people still represents a challenge due to definitional problems in identifying the indigenous group and the heterogeneity of concepts and terminology used by different countries. Notwithstanding the above mentioned constraints, the demand for statistics in disaggregated forms is on the increase. In order to realistically meet this demand the following actions may be useful:

i. Standardize concepts and classifications pertaining to indigenous issues so that they are statistically operational and a basis for collecting reliable and valid data.

ii. Maximise the use of different sources of data on indigenous people, namely, censuses, surveys and administrative record systems.

iii. Efforts should be made to disaggregate data on indigenous people by sex and other sub-group classifications of the populations.

References

Alemany, J and Zewoldi, Y. (2003). Ethnicity: Data Collection and Dissemination. Paper prepared for the United Nations Statistics Division.

Aspinall, P. (2000). Operationalising the collection of ethnicity data in studies of the sociology of health and illness, Sociology of Health & Illness, 2000, 23, 6, 829-862.

Edwards, R. W. (2003). ABS Views on Content and Procedures. Australian Bureau of Statistics, Canberra, Australia, July 2003.

Evinger, S. (1995). How Shall We Measure Our Nation’s Diversity? Black, White, Cape Verdean, German-American, Or? Washington reviews how to define the population for the 2000 census. Springer-Verlag, New York. Inc. pp.1-18

______(1996). How to Record. America Demographics, May 1996, pp.36-41.

Hungarian Central Statistical Office. (2002). Population Census 2001: Ethnic affiliation. Prepared by the Population Census Department from the Hungarian Central Statistical Office, 2002.

Tishkov, V. (2001) Ethnicity and Language in Russia’s 2002 Census. Paper presented in the International Symposium “Population Census – XXIst Century: Experience – Problems – Prospects”, Moscow, Russian Federation, 27-28 November 2001.

United Nations (1956) Demographic Yearbook. United Nations Publication, New York.

(1963) Demographic Yearbook. United Nations Publication, New York. (1964) Demographic Yearbook. United Nations Publication, New York.

(1971) Demographic Yearbook. United Nations Publication, New York.

(1973) Demographic Yearbook. United Nations Publication, New York.

(1979) Demographic Yearbook. United Nations Publication, New York.

(1983) Demographic Yearbook. United Nations Publication, New York.

(1988) Demographic Yearbook. United Nations Publication, New York.

(1993) Demographic Yearbook. United Nations Publication, New York.

(1998). Principles and recommendations for Population and Housing Censuses. Statistical Papers, No.67, Sales No E.98.XVII.8.

(2002). Permanent Forum on Indigenous Issues, Report on first session (13-24 May 2002), E/CN.19/2002/1/ Rev.1

Waters, M. (2000) Immigration, Intermarriage, and the Challenges of Measuring Racial/Ethnic Identities, American Journal of Public Health, November 2000, 90, 11, pp. 173-175.

1 The views expressed in this paper are those of the author and do not imply the expression of any opinion on the part of the United Nations Secretariat. 2 These countries were: Mozambique, South Africa, and Zimbabwe. Please note that they are all part of SADC (South African Development Community) and have agreed to ask similar questions in their census.

3 The Russian Federation, for example, recognized as many as 191 nationalities