Abstract
The author observed the conduct of the national census at an outstation community in the Northern Territory.
The first three sections of the chapter describe in some detail the recruitment and training of the enumerators and the enumeration process itself. The fourth section argues that as a head count, the enumeration strategy was as successful as it could be under the circumstances of high population mobility and the long period of time involved in the enumeration. In the final sections the author examines the quality of the data in terms of household structure and kinship. It is argued that it is impossible to elicit information about one kinship system in terms of another, and that the resulting data is doomed to be uninterpretable.
The author concludes that the effectiveness of the IES can be measured according to the effectiveness of both the recruitment and training of the Indigenous enumerators and the strategies for eliminating under-enumeration and double counting. There are several suggested changes for the form and content of the SIPF which would result in a shorter and more 'culture-specific' form and a higher proportion of self-administered forms, resulting in the collection of more reliable and relevant data.
In August 2001 I observed the conduct of the national Census at an outstation community (henceforth community A), serviced by a homelands association (henceforth HA) which is based at the Aboriginal settlement of B in the Northern Territory. The purposes of this research were twofold: to evaluate the IES as it was applied in this particular context, and to assess the quality of the data that were collected.
This study is being published under terms of the Census and Statistics Act 1905 Undertaking of Fidelity and Secrecy, which imposes strong principles of confidentiality to protect the anonymity of census respondents. Because community A is small and its residents would be readily identifiable (at least locally) if it were named, the area in which this case study took place is left deliberately vague, and communities are referred to by letters of the alphabet rather than by names. The local Indigenous language is not referred to by name, and where Indigenous language terms occurred in the quotations cited, they are replaced by an English translation in square brackets. The kinship terms of the local system have been replaced by the short-hand forms conventionally used in anthropological publications (see explanation of kinship terminology and abbreviations on p.ix ). None of the actors in this story are identified by name: officers of the census are referred to by an acronym for their function, and other individuals are referred to by a letter, or as 'person 1' in a household (where applicable). The dwellings are not individually identified by number. The data on the composition of the study community and of its households are described as if they were actual examples, in order to avoid tortuous syntax, but they should be understood for what they are: artificially constructed data sets that do not correspond to the data on the census forms for the community. The examples are hypothetical, but constructed to bring out the structures and issues that emerged clearly from the actual census data.
The first three major sections of the chapter describe in some detail the recruitment and training of the enumerators and the enumeration process itself. The intention is to provide an ethnographic account of the process in its context, as a background to the main findings and recommendations. Although the IES has been in existence in various forms since the 1976 Census (see Taylor, this volume), there is a dearth of detailed research on its implementation on the ground, and on its effectiveness as a strategy for obtaining complete and accurate information on Australia's Indigenous population.
David Martin and John Taylor, in their observation and analysis of the 1986 and 1991 Census enumerations at the remote community of Aurukun on Cape York Peninsula, found a systematic under-enumeration of children and young adults, and identified several- 30 - contributing factors, most significantly the mismatch between standard definitions of the 'household' and the 'family' and the facts of Aboriginal social organisation on the ground, and the relatively high mobility of Aboriginal people, both between households and within particular regions (Martin & Taylor 1995: 13, 14–17).
It will be argued in the fourth section that, considered purely as a head count, the enumeration strategy employed at community A in 2001 was largely successful. The reasons for its success are discussed: it is important to understand the reasons for success as well as for failure.
The quality of the data collected is another matter. Particular attention is given, in the fifth part of the chapter, to continuing problems with the definition of 'household' and 'family', and with the questions on the two forms—the SIHF and the SIPF—that compounded the problem of collecting reliable and meaningful data on household structure. In the final section, the quality of the other data is examined.
At a more general level, in these final two sections of the chapter, consideration is given to the kind of information that the ABS is attempting to collect in the census, and whether the design of the forms and of the questions they contain could be better tailored both to the intended respondents and to the goals of the census. The chapter concludes with some general recommendations that flow from the data and the analysis.
The rough estimate for the current Indigenous population of the SLA in which community A is located, derived from the Community Housing and Infrastructure Needs Survey (CHINS) that was conducted immediately prior to the 2001 Census, is 10 000 people.[1] All of this area is remote, and most of its Aboriginal inhabitants are 'traditionally oriented' (to use a convenient shorthand term). The major communities are accessible by reasonable roads in the dry season, but many outstations can be reached only by many kilometres of rough track, or by light aircraft.
The CFO responsible for the enumeration in this SLA planned for a rolling census, with different communities being enumerated at different times, starting in the second half of July and ending by census day itself. This strategy was dictated by the logistics of training the CCs and local Indigenous interviewer-collectors (henceforth enumerators), and processing the results of the enumeration. The schedule was badly disrupted by the deaths and subsequent funerals of two people at two different communities within the SLA. These events delayed or halted the training of the enumerators and the enumeration itself. This was unfortunate for the CFO, for whom it created something of a logistical nightmare.
As will be shown in more detail below, the enumeration in this region was carried out on the basis of persons actually present at a dwelling at the time of the enumeration; that is, according to the standard census procedure. I will argue that if it is consistently administered by enumerators with detailed local knowledge, this strategy is less inherently error-prone and more amenable to verification than the alternative strategy adopted in other areas of the Northern Territory, that of a 'usual residents' count.
- 31 -The population of community A fluctuates from around 20 (when there is a large ceremony taking place elsewhere) to over 250 (when there is a ceremony taking place at the community); its normally resident population is in the region of 100 people. At the time of the enumeration about 50 per cent of the people listed as 'living here' on the SIHFs were away, at a location where a major funeral that had just taken place, or visiting relatives living elsewhere in the area. But the community was also host to a number of visiting relatives from the immediate region, and from three other communities (C, D and E) outside the immediate region.
The community is structured around the members of two related patrilineages (lineage X and lineage Y) on whose land the community lies. X and Y, the men from whom these two lineages are descended, are classified as brothers. Fig. 3.1 shows the kin relationships, in simplified form, of the people designated as 'person 1' for each occupied dwelling. The census does not capture (because it does not seek to) these inter-household relationships, but it is arguable that the data collected by the census on the composition of individual households make sense only in the context of this larger picture.
Figure 3.1. Kin relationships between people designated as 'person 1' for each occupied dwelling, community A, Census 2001
At the time of the census there were 13 occupied dwellings in the community. The census enumeration took place over three days, between 6 and 8 August. During this period the community was preparing for the funeral of an infant who had died the previous week, and it received a visit from a candidate for one of the two major political parties, canvassing for votes in the upcoming Northern Territory election.
- 32 -[1] CHINS provides an estimate of the service population rather than a census-style head count. CHINS estimates invariably, therefore, produce larger populations than does the census enumeration.