PFII/2004/WS.1/9

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)

Collection and Disaggregation of Data Concerning Indigenous Peoples: An Australian Perspective

This paper discusses the on-going challenges to our understanding of data collection and disaggregation on Indigenous peoples. The discussion is framed around the agenda of the upcoming workshop and while the paper’s examples are Australian, the issues highlighted are common to Indigenous peoples worldwide.

1.0 Introduction

Reliable data on Indigenous peoples are a vital pre-cursor to effective policy. Such data are essential for setting policy parameters, facilitating effective decision-making and resource allocation, evaluating policy, as well as assessing the on-going social, economic, cultural well-being of Indigenous peoples. Despite this critical need, Indigenous data collection is a relatively recent phenomenon in Australia. Indeed, until amended by referendum in 1967, Section 127 of the Australian Constitution specifically excluded the ‘aboriginal race’ from official population figures (Chesterman & Galligan 1997). Since the 1970s, a coordinated approach to Indigenous data collection and disaggregation has developed only slowly and collection systems remain incomplete and unreliable. Therefore, while the extremely disadvantaged socio-economic position of Australian Indigenous peoples are obvious, the lack of comprehensive, historical Indigenous data continue to hinder policy formulation and evaluation. Without such data it is difficult to demonstrate either the effects of government policy or to differentiate these from other macroeconomic factors. This paucity of policy-relevant Indigenous data was highlighted by the1991 Royal Commission into Aboriginal Deaths in Custody (RCIADIC). Linking low socio-economic status with high rates of incarceration among Indigenous peoples, the Commission recommended an improved data collection to allow a more refined analysis of the causes of social marginalisation (Altman 1996).

2.0 Who are the Indigenous people of Australia?

The Australian Indigenous population is estimated at around 458500 persons, or 2.4 per cent of the total population. Clearly identifiable differences exist between the Indigenous and non-Indigenous populations. The Indigenous population is younger; more likely to reside in areas classified as very remote; and is disproportionately disadvantaged across health and socio-economic indicators (ABS 2002). The Indigenous peoples of Australia are comprised of two separate groups, Australian Aborigines and people from the Torres Strait Islands. Of the Indigenous population, Torres Strait Islander people making up about 10.6 per cent of the total (ABS 2001). While conventionally combined for statistical purposes the two populations are not homogeneous and have significant demographic, political and cultural differences.

The generally used Australian definition states that an Indigenous person is one who:

i. has Aboriginal or Torres Strait Islander descent

ii. identifies as an Aboriginal or Torres Strait Islander person

iii. is accepted as an Aboriginal or Torres Strait Islander by the community in which he or she lives.

This definition is widely accepted within Government departments, statutory bodies and institutions, but sometimes contested as imposed by Aboriginal groups. Based around this definition, in 1996, the Australian Bureau of Statistics (ABS) adopted a standard question for Indigenous status.[1] This innovation is regarded by the ABS as a considerable advance as it provides consistency over time and over data collections and allows valid ratio and rate statistics to be produced (Barnes 1996).

3.0 Major Sources and Methods of Data Collection

The major source of data collection on Indigenous Australia remains the 5-yearly national census. The ‘whole of population’ nature of the data collection, the similarity of the data collected from year to year, and the direct comparability of the data, places this collection as the key source of Indigenous statistics. However, the scope of these data are necessarily limited and the 5-yearly collection cycles are too long for timely policy analysis. Also, as Altman and Taylor (1996) note, the census is essentially a vehicle for the collection of information on all Australians and the Indigenous data amassed is merely the ‘by-product of including an Indigenous identifier’ (p. 193). Thus, within the census there is no capacity to accommodate the cultural social or economic differences between Indigenous and non-Indigenous Australians, or to address the specific data needs of the Indigenous population.

The other major sources of Indigenous data are the administrative data collections from places such as hospitals, housing and welfare agencies and criminal justice systems. In very recent years the ABS has implemented strategies to improve the quality of administrative Indigenous data. For example, the Registry of Births, Deaths and Marriages has now adopted the ABS standard Indigenous identifier in their birth and death collections in all States. However, while there are signs of improvement, data from administrative sources remains incomplete and prone to inaccuracy, with the haphazard nature of recording clients’ Indigenous status a key area of concern.

The governmental response to the RCIADIC identified deficiencies in Indigenous data collection was the National Aboriginal and Torres Strait Islander Survey (NATSIS). Conducted in 1994, this survey was the first national specific Indigenous data collection exercise in Australia. While a major purpose of NATSIS was to establish base line data for future comparison, the NATSIS has not been repeated. Instead, in 2002, the ABS conducted an Indigenous Social Survey (ISS) with the first results due for release in early 2004. Discussions with the ABS indicate that data from this survey will be broadly comparable with NATSIS data and will build on the data collected there (NCATSIS 2003). The ABS plans to run an ISS every six years. However, while a regular ISS is a much-needed initiative, the availability of reliable and comparable Indigenous data remain a future aspiration, not a current reality.

4.0 Major challenges

Apart from a lack of reliable data, three other challenges to the collection, disaggregation and dissemination of Indigenous data can be clearly identified.

4.1 Comparability of Data

The direct comparability of intra-national Indigenous and non-Indigenous data remains a challenge for a number of reasons. The first is differences in question structure between Indigenous specific and mainstream surveys. For example, Anderson and Sibthorpe (1996) using NATSIS and the National Health Survey (NHS) data on health care seeking behaviour find that in no case is the wording of the questions in the two surveys exactly the same. This disparity has been partly addressed in later NHS surveys via a supplemented Indigenous sample, but the range of estimates produced is restricted and available at the national level only (ABS 2003). Second, large-scale demographical differences pose a comparability challenge. For example, while most data indicates that the unemployment rate of the Indigenous population runs at least twice that of the non-Indigenous population, how much of this difference can be attributed to differing age structures? Adjusting for age in Indigenous populations around unemployment has been shown to lead to a still significant but less dramatic difference (Jackson 2002). Age adjustment of health data works in the opposite direction. While initial comparison of data indicate that Indigenous rates of cancer are lower than those of the non-indigenous population, adjusting for age finds the opposite (ABS 2001). Third, data comparability can be compromised by methodology and/or cultural interpretations and definitions. For example, while the household is a standard comparison category, is an Indigenous household one where the reference person or spouse is Indigenous (as per ABS census results), or one where any resident is Indigenous (as per the NATSIS collection)? The differing structure of many Indigenous households from the non-Indigenous norm also presents challenges of comparability and appropriateness of standard definitions for Indigenous populations (Altman 1996). Finally, given the above, the scope for inter-country Indigenous data comparisons is extremely limited.

4.2 ‘Small Percentagisation’

Another major challenge can be termed ‘small percentagisation’. The basic issue is that a proportionally small Indigenous population (2.4%) means that the Indigenous sample of even a large-scale survey is too small to generate reliable results. Because of this, many previous national surveys, such as the ABS Survey of Families (1992), did not include an Indigenous identifier (Taylor 1996). While Indigenous results are increasingly reported in very recent major surveys, small samples mean that they tend to be disaggregated to the national level only, with restricted variable reporting. Non-ABS surveys are also problematic. Several longitudinal social surveys have recently been inaugurated in Australia. Of these, the Indigenous sample in the Household, Income and Labour Dynamics in Australia (HILDA) survey, (259 cases from a total of 13969) is too small for significant analysis and the Longitudinal Study of Australian Children begun in 2003 will also only include those few Indigenous children who emerge from their random sampling process (AIFS 2002). The Negotiating the Lifecourse (NLC) Survey does not include an Indigenous identifier.

While separate surveys such as the NATSIS or ISS overcome some of the issues, they are not miracle cures. Apart from the risk of low data comparability, national Indigenous specific surveys can divert attention from the responsibilities of other agencies, risk becoming omnibus surveys in which limited data about too many issues are collected, and are prone to ideological and political influences on the accessibility, type and level of data collected and released (Altman 1996; Carcach &Mukherjee 1996). Yet, the counter argument of enhanced or weighted Indigenous samples is also problematic: costs are high; they can impose a high response load on Indigenous populations; and risk minimising the significant regional, demographic and socio-economic differences within the Indigenous population.

4.3 Data Accessibility

Another major challenge is the accessibility of what Indigenous data are collected. Non-ABS researchers, Indigenous and non-Indigenous, generally have to rely on statistics and tables produced by the ABS or the Australian Institute of Health and Welfare (AIHW). While the ABS does release data as Confidential Unit Records Files (CURF), the number of datasets made available through this process is restricted. Also, with CURF datasets, the researcher can only utilise the data in accordance with the coding decisions made by the ABS. For example, the NATSIS data released in CURF format did not include the remote area identifier, had some housing tenure variables excluded, and had variables such as hours worked and individual income grouped into broad categories. As Jones (1996) notes, while such decisions are no doubt based on justified concerns to protect individual confidentiality, such limitations severely limited researchers ability to use the NATSIS to investigate Indigenous poverty, a primary reason for its conduct in the first place. My own experience in attempting to use NATSIS data as a baseline comparison for a 2003 study on the housing circumstances of Tasmanian Aborigines was similarly confounded by the exclusion of the state-identifying variable. The only alternative was to purchase sets of cross-tabulated data direct from the ABS; an expensive and second-best option. The ISS, according to the ABS, will overcome some confidentiality concerns by using a larger Indigenous sample (ABS 2000).

5.0 Stakeholders

The stakeholders in the collection, production, dissemination and usage of Indigenous data are manyfold. The sheer number of stakeholders is, in itself, a challenge to Indigenous data collection. As Altman (1996) argues, the diversity of views about the appropriate balance between types and content of data collection, data integrity, analytic amenability and policy worthiness risk the development of too many, and potentially clashing, data demands. Even limiting the list to Indigenous stakeholders produces a very broad grouping. In Australia, these include peak Indigenous bodies such as ATSIC regional councils and policy staff, local Indigenous communities and their representative bodies, representatives of professional and interest groups such as the National Aboriginal Community Controlled Health Organisations (NACCHO), Indigenous policy makers, and academics, professionals and researchers based in tertiary or bureaucratic institutions.

5.1 Stakeholders’ Data Needs

While Indigenous stakeholders share a common interest in the production of valid, usable Indigenous data, communities, organisations and institutions have differing information needs. As a result, Indigenous stakeholders tend to hold divergent perspectives on the type of data collected, collection methods and use of this data. For some stakeholders data that lend themselves to international comparisons and informing national debates are the most important, while others stipulate that data collection should have regional interests as their focus, In turn, peak Indigenous bodies and policy makers have their own distinctive organisational or institutional requirements (Altman 1992). For example, Anderson and Sibthorpe (1996) note that ATSIC regional councils require data useable at the regional council level for planning purposes, while the information needs of community organisations tend to be shaped by the concerns of community members and local program requirements. In particular, the Indigenous population heterogeneity means that data at the national level only are frequently of limited assistance for accurate evaluation of Indigenous circumstances and may actually obscure major differences in Aboriginal populations. At the minimum, to be of use to a significant proportion of Indigenous stakeholders, data needs to be disaggregated to the ATSIC regional council level (35 in total). However, as noted, national level data are generally all that is currently available.

5.2 Current Involvement of Indigenous Stakeholders

In line with other Indigenous peoples, Indigenous Australians frequently articulate concerns about the ethics of data collection processes and the impact of these on those who are the subjects of research. There exists a longstanding, widespread, and in many cases, justifiable suspicion within Indigenous communities about data collection. Within my own community, research proposals often receive a cynical response and strong questioning relating to community value and benefits. Unfortunately, such concerns are not always easy to allay.

This general mistrust is not aided by the fact that the collection, analysis and dissemination of Indigenous data in Australia remain a largely Indigenous-free zone. Although Indigenous people are employed in the collection of census and survey data, the proportion of Indigenous researchers, policy makers or other empowered stakeholders is few. For example while the ABS maintains a National Centre for Aboriginal and Torres Strait Islander Statistics (NCATSIS), the Centre currently has no Indigenous staff (NCATSIS 2003). Of those that are involved, fewer still are instigators. While Indigenous consultation is now ABS policy, Indigenous representatives, such as those on reference committees, tend to be in the minority with their selection at the discretion of the non-Indigenous bureaucracy. Therefore, while significant attempts have been made by the ABS to include more Indigenous input into the process of collecting Indigenous data (ABS 2003), perceptions of tokenism among Indigenous stakeholders remain strong. The responsibility placed on these Indigenous representatives to represent ‘the Indigenous perspective’ is also an onerous one. Regardless, decisions about how Indigenous data collection is carried out, what data are collected, how data are analysed and disseminated and the policy initiatives emanating from data are still controlled by a non-Indigenous bureaucracy.

5.3 Appropriate Involvement of Indigenous Stakeholders

Suspicion of data collection processes can translate into suspicion about the value and usefulness of Indigenous data. In particular, there is a reluctance by many Indigenous groups to use statistical or ‘hard’ data for planning purposes or to engage in research themselves. This reluctance is manifested in unease around data validity and usefulness, or the rejection of the use of such data as ‘not culturally appropriate’. From these perspectives, statistical data are deemed irrelevant to the imperatives of the Indigenous political, social and economic domain. For example, a survey of ATSIC offices on the NATSIS (1994) regional statistics, found that over 70 per cent felt that the data were not useful to them because they did not use socio-economic data in regional planning, or because the data were not viewed as culturally appropriate (Martin 1996). A general lack of skills within Indigenous organisations to make effective use of statistical data compounds this reluctance.

The place of an Indigenous agenda and priorities in current data collection planning is also questionable. While the ISS asks a limited series of questions relating to culture and language, their relevance or level of importance to Indigenous people is unclear. Critically, there is little evidence to indicate serious contemplation of the notion that Indigenous priorities on what data ought be collected may not correlate with the priorities of the major data collection agencies or the Australian State. Yet, while the ABS and its international counterparts may be most interested in demographic, health or socio-economic change, for Indigenous peoples, indexes of self-determination or cultural maintenance may be of greater or equal importance. The statistical requirements of data also pose risks. The need for the collection of robust data or the compiling of easily comparable indexes can sit at odds with the developing data around complex issues such as cultural values or identity. The difficulty of collecting, measuring and comparing such data may mean they are deleted or excised from collections for statistical reasons. The question of whether surveys are the most appropriate way to collect such data is also a valid one.

The influence of culture and cultural values on the validity and usefulness of data to both Indigenous and non-Indigenous stakeholders also appears underplayed. A clear and important example is health. The connotations and meanings of various concepts used in data collection around health have cultural implications. As Anderson and Sibthorpe (1996) point out, health is a culturally dense concept, in both the Aboriginal and non-Aboriginal context. As these authors argue, depending on the cultural context ‘healthy’ might imply an absence of acute distress, being personally mobile and non-institutional or not having trouble with kin or community relations. There is also considerable cultural variation within the Indigenous populations, and although such factors complicate the gathering and analysis of data, they need to be recognised and addressed. Indigenous communities’ weak connection to the importance of data collection is currently being addressed by the ABS. The upcoming implementation of an Indigenous Community Engagement Strategy will focus on enhancing the involvement of Indigenous communities at the regional level and provide community training (ABS 2003). However, in the present, this initiative, along with others, such as the ISS, is untried or only very recently implemented.

Overall, therefore, Australian Indigenous peoples remain largely marginalised from the collection, production and use of Indigenous data. As a result, providing detailed comment on the appropriate involvement of indigenous experts and organisations in planning development, analysis and dissemination of statistical work is a difficult exercise. While Indigenous participation at all levels of the process is obviously both desirable and functional, in the current situation, an articulation of appropriate Indigenous involvement would be merely a speculative wish list. Additionally, while the recent efforts of the ABS, as the major collector and disseminator of Indigenous data in Australia, are acknowledged, the institutional understanding of Aboriginal interests, perspectives and life experiences still appears to be very limited. The implementation of strategies such as those designed to engage Indigenous communities in data collection are important steps forward, but, as yet, there appears little recognition that such engagement is a two way process.

6.0 Collecting Intra- And Inter-Country Comparable Data

This issue of intra-country collection has largely been discussed within the previous items. Currently, Australian Indigenous data collection is limited by: the extent to which Indigenous people are included in national surveys; the accuracy with which Indigenous people are identified in both surveys and administrative datasets; uncertainties about Indigenous population estimates; and concerns about whether the survey methods employed are always the most suitable method of data collection However, as also noted, considerable work has been undertaken in very recent years towards improving and standardising methods of data collection and analysis, and increasing the cross-comparability of data between the Indigenous and non-Indigenous populations.

Cross-national comparisons are more difficult again. The identified deficiencies in the collection and disaggregation of Australian Indigenous data appear replicated in various degrees of severity in other nations with Indigenous minorities such as Canada. Classifications of who is and who is not an Indigenous person also vary significantly across nations. Yet, cross-national comparability of Indigenous data is important. Such comparison can: provide benefits in data analysis; allow the development of international yardsticks on national progress in overcoming Indigenous disadvantage; point to instances of successful policy as well as identifying negative trends; plus allow Indigenous people to clearly compare their own circumstances with their brothers and sisters in other nations. For example, at a recent International Indigenous Health Knowledges conference, delegates from New Zealand, Australia, Canada and the United States called for the development of an internationally comparative report card on the progress of individual national governments in addressing disparities in Indigenous health. Such a strategy obviously requires internationally comparative health data.

Given the state-based control of major data collection sources, commitment from governments on the desirability and facilitation of international data comparability is a non- negotiable pre-cursor. A starting point, therefore, would be an international agreement between core national data collection agencies on facilitating the comparability of Indigenous data. Crucially, Indigenous stakeholders would need to be integrally involved in this process to ensure that Indigenous perspectives, worldviews, knowledges and agendas are incorporated from the start into such data collection.

References

ABS (2000), Directions in Australia's Aboriginal and Torres Strait Islander Statistics, Australian Bureau of Statistics, Canberra.

ABS (2001), The Health and Welfare of Australia's Aboriginal and Torres Strait Islander Peoples, Cat. No. 4704.0, Australian Bureau of Statistics, Canberra.

ABS (2003), The Health and Welfare of Australia's Aboriginal and Torres Strait Islander Peoples, Cat. No. 4704.0, Australian Bureau of Statistics, Canberra.

AIFS (2002), Introducing the Longitudinal Study of Australian Children, LSAC Discussion Paper No. 1, Australian Institute of Family Studies, Melbourne.

Altman, J. C. (1992), Statistics about Indigenous Australians: needs, problems, options and implications, A National Survey of Indigenous Australians: Options and implications, J. C. Altman, Centre for Aboriginal Economic Policy Research, The Australian National University, Research Monograph No. 3, Canberra: 193-202.

Altman, J. C. (1996), The 1994 NATSIS: From conception to completion and beyond, The 1994 National Aboriginal and Torres Strait Islander Survey: Findings and Future Prospects, J. Altman & J. Taylor, Centre for Aboriginal Economic Policy Research, The Australian National University, Research Monograph No. 11, Canberra: 1-12.

Altman, J. C. & Taylor, J. (1996), Statistical needs in Indigenous affairs: Future options and implications, The 1994 National Aboriginal and Torres Strait Islander Survey: Findings and Future Prospects, J. Altman & J. Taylor, Centre for Aboriginal Economic Policy Research, The Australian National University, Research Monograph No. 11, Canberra: 193-202.

Anderson, I. & Sibthorpe, B. (1996), The NATSIS and Policy and Planning in Aboriginal Health, The 1994 National Aboriginal and Torres Strait Islander Survey: Findings and Future Prospects, J. Altman & J. Taylor, Centre for Aboriginal Economic Policy Research, The Australian National University, Research Monograph No. 11, Canberra: 118-134.

Barnes, A. (1996), Recent Developments in the Collection of Indigenous Statistics, The 1994 National Aboriginal and Torres Strait Islander Survey: Findings and Future Prospects, J. Altman & J. Taylor, Centre for Aboriginal Economic Policy Research, The Australian National University, Research Monograph No. 11, Canberra: 13-27.

Carcach, C. J. & Mukherjee, S. K. (1996), Law, justice, Indigenous Australians and the NATSIS: policy relevance and statistical needs, The 1994 National Aboriginal and Torres Strait Islander Survey: Findings and Future Prospects, J. Altman & J. Taylor, Centre for Aboriginal Economic Policy Research, The Australian National University, Research Monograph No. 11, Canberra: 156-164.

Chesterman, J. & Galligan, B. (1997), Citizens Without Rights, Aborigines and Australian Citizenship, Cambridge University Press, Cambridge.

Jackson, N. O. (2002), 'The Doubly-Structural Nature of Indigenous Disadvantage. Indigenous Age Structures and the Notion of Disparate Impact', New Zealand Population Review, 28,(1), 55-68.

Martin, D. F. (1996), The NATSIS as a regional planning and policy tool, The 1994 National Aboriginal and Torres Strait Islander Survey: Findings and Future Prospects, J. Altman & J. Taylor, Centre for Aboriginal Economic Policy Research, The Australian National University, Canberra: 173-182.

NCATSIS (2003), National Centre for Aboriginal and Torres Strait Islander Statistics, Australian Bureau of Statistics, Darwin.

Royal Commission into Aboriginal Deaths in Custody (1991), Recommendations, Australian Government Publishing Service, Canberra

---

[1] In 1995 the ABS formally adopted the following question as the standard for identifying persons as members of the Indigenous population.

Are you of Aboriginal or Torres Strait Islander origin?

For persons of both Aboriginal and Torres Strait Islander origin please mark both ‘Yes’ boxes. Response options are: No; Yes, Aboriginal; and Yes, Torres Strait Islander