This monograph explores some of the problems, successes and policy issues related to the application of the Indigenous Enumeration Strategy (IES) in the enumeration of Aboriginal people in remote parts of Australia. It is based on the evidence of direct observations made by three researchers from the Centre for Aboriginal Economic Policy Research (CAEPR) of the conduct of the 2001 Census enumeration in three separate localities. The localities—a major Aboriginal township, an outstation, and a series of urban town camps—were deliberately selected to be broadly representative of Indigenous settlement patterns across the remote north and centre. The aim was to sample a range of residential settings in the event that this variation had any bearing on the conduct and outcomes of the census. Access to the enumeration process was facilitated under the terms and conditions of confidentiality as specified by the Census and Statistics Act 1905. Each researcher was a signatory, thereby assuming official census 'observer' status.

Prior to the 1971 Census, relatively few resources were applied to the enumeration of Indigenous people resident in isolated localities, and the focus was very much on achieving a head count rather than a detailed profile of individual social and economic characteristics. Since that time, special census field procedures have been progressively devised, modified and extended by the Australian Bureau of Statistics (ABS) in an attempt to ensure as comprehensive coverage as possible of remote area populations, albeit within budgetary constraints. The direct cost of enumerating remote area Indigenous populations in 2001 was around $2 million (or $26 per head of population) compared with the direct cost of around $49 million (roughly $2.60 per head) for the total population. This cost excludes a number of associated costs such as form printing and transport, the processing of forms and the preparation of outputs (pers. comm. Paul Williams, ABS).

There are difficulties, however, in determining how effectively these monies are deployed in the pursuit of an optimal count and responses of good quality. The normal method of checking for accuracy in the count, using the post-enumeration survey, is not applied to the remote area Indigenous population. There is no standard test for the quality of the data on population characteristics, short of editing rules that might be applied at the data processing stage.

In remote Aboriginal communities, then, there is a need for close scrutiny of census procedures to assess whether existing methods produce optimal results, or whether alternative or refined methods are needed. These assessments, in turn, have resource implications of interest beyond the confines of ABS operational systems. For example, the Aboriginal and Torres Strait Islander Commission (ATSIC) considers that both planning and policy formulation begin with demographic facts. Indeed, the broad parameters of ATSIC's charter are determined by the size, growth, composition and changing location of the Indigenous population. These factors also provide the basis for assessing issues of social justice such as the recognition of need and the fair and equitable distribution of resources (Menham 1992: 37).

- 2 -

The ABS has claimed at times, on the basis of its own qualitative assessment, that the enumeration of remote area Indigenous populations may actually produce an overcount (ABS 1993: 6), yet analysts and other users of remote area census data have often asserted that the enumeration sometimes underestimates the numbers of Indigenous people. One particularly forceful claim, reported as part of a joint initiative of the Queensland and Commonwealth governments in the form of the Cape York Peninsula Land Use Strategy (CYPLUS), was that Indigenous people in Cape York Peninsula were substantially undercounted in 1991 (King 1994). However, this assertion was based on an invalid comparison between data from the 1991 ABS Census de facto counts of the Cape York population, and estimates of the 1994 place of usual residence (de jure) population derived from a variety of key informants in Cape York communities. To use the author's own assessment, it was constructed from deductive guesswork based on questionable assumptions (King 1994: 27–8). Such assertions of underenumeration provide no statistical basis for testing the proposition and serve only to obscure the likely underlying causes. Indeed, the validity of such contrary claims is difficult, if not impossible, to establish in the absence of demographic data that are directly comparable to those collected by the census.

It has also been claimed that the nature of census questions and the respondents' interpretations of them may misrepresent personal characteristics and patterns of social and economic organisation in remote Aboriginal communities (Ellanna et. al. 1988: 193–7; Commonwealth of Australia 1992; Jonas 1992; Martin & Taylor 1996; Smith 1992). Misrepresentation of social and economic characteristics can occur because the concepts underpinning the census questions lack cross-cultural fit. Data on income provide a good example. The census measure of income applied in remote Indigenous communities refers to a period of time—a typical fortnight—whereas the flow of income to individuals and households in remote Indigenous communities is often intermittent. It is difficult to determine what might constitute usual fortnightly income in many Aboriginal households. Intermittent employment and windfall gains from sources such as gambling, cash loans and royalty payments, combined with debits, for example due to loss of employment and cash transfers to others, create a highly complex picture even over a short space of time—one that census methods of data gathering are likely to misrepresent (Smith 1991).

Attempts to describe household composition provide another example. In census terms, a household is defined as a group of two or more persons, who usually reside in the same dwelling, who regard themselves as a household, and who make common provision for food or other essentials for living. Visitors to the household are not included. Ethnographic evidence suggests that these are highly problematic definitions when applied to Aboriginal households, particularly in remote areas.

While there is some utility in confining the notion of a household to residents of a physical dwelling or location, Aboriginal households are typically highly fluid in composition, often with a more or less stable core of residents and a variable periphery of transient residents drawn from the same community or regional population pool. In such circumstances, it is clear that co-residential groupings (even in the limited sense of who sleeps where), commensal units, family groupings, and domestic economic units are not necessarily coterminous; for example, people who live together may not eat together.- 3 - Commonly too, the basic economic and social units of Aboriginal societies comprise linked rather than single households (Altman 1987; Finlayson 1991; Henry & Daly 2001; Martin & Taylor 1996; Smith 1991, 1992), and what Aboriginal people themselves refer to as 'families' are typically dispersed across a number of households. It is such clusters of households, rather than individual households, which commonly form the basic units of sociality and consumption in remote Aboriginal communities.

The accuracy of census counts and of data on population characteristics, cannot adequately be established without reference to the particular circumstances of census-taking in remote Aboriginal communities. To date, a basic problem with the interpretation of census results has been the absence of well documented and fully nuanced accounts of this encounter between communities and officialdom.

There are some precedents for independent scrutiny of ABS field procedures. In 1981, two academics were invited to observe and report on preparations for the census count of remote Aboriginal communities in the Northern Territory (Loveday & Wade-Marshall 1985). This exercise focused on three aspects of these preparations: the organisation of field procedures and training of collectors, the pre-census publicity campaign, and issues likely to arise from the administration of census forms and questions. All information was gleaned from meetings convened by Census Field Officers (CFOs). The actual count was not observed.

While scrutiny of the preparations for the 2001 remote Indigenous census count also formed part of the present exercise, the intention was much more to focus on the conduct of the count at the community level, as well as on the process of house to house interviewing. The framework, then, is a critical assessment of the application of western census-taking methods in a cross-cultural context. Loveday and Wade-Marshall (1985: 249) found that preparations for the 1981 Census threw into sharp relief the difficulties encountered by the ABS in translating census questions into the context of Aboriginal society in various kinds of settlement and in remote outstation locations. For the 2001 Census, the ABS has acknowledged a need for feedback on census field operations, as well as a desire to be more informed about the social and cultural contexts in which their efforts are expended. This has been some time in coming, although as with all shifts in administrative process, it has an essential history.