The effectiveness of the IES can be measured according to several criteria. The strategy for the recruitment of the Indigenous enumerators appears to have been effective, particularly when one bears in mind the disruptions that occurred. The strategies for eliminating under-enumeration and double counting were about as good as they could be, in circumstances that are inherently extremely difficult. Some fine-tuning in these areas is probably possible and desirable, but radical rethinking is not required.
Because I was not able to observe a training session, and because the training of the enumerators at community A was curtailed, it is difficult to comment directly on training procedures and content, and their effects. In the recommendations below I suggest quite radical changes to the form and content of the SIPF. If implemented, the training of the CFOs and enumerators would need to change in any case, and therefore:
I would strongly recommend that any new version of the SIHF and/or SIPF be pre-tested in the field in a range of communities, using as enumerators people who were employed as enumerators in the 2001 Census. They, in conjunction with a CFO who had also taken part in the 2001 Census exercise and an anthropologically-trained observer would be in the best position, between them, to evaluate the form and make recommendations for any further refinements. Such an exercise could also act as a pilot for new training procedures. Such training should focus as much on the contents of the questions on the SIHF/SIPF as on procedural matters.
Detailed recommendations concerning the form and content of the questions have been made throughout the second half of this paper. The following, more general, recommendations flow from them:
The SIPF is much too long. Administering it was an onerous task for the enumerators, and this encouraged formulaic responses to certain questions, particularly ones whose purpose was opaque to local people, or which were worded or structured in ways that made them hard to interpret.
The presence of questions on kinship at the beginning of the SIPF, and the fact that they also appeared on the SIHF, gave them undue emphasis. Because kinship is such a salient topic for local people, they took this as a signal that these were the most important questions to address, and more time was spent on Q.4, and on thinking about how it should be answered, than on any other question. I have suggested that the idiom- 72 -of kinship be dispensed with in eliciting information on household composition, and that a new type of extended family household be added to the inventory of ABS household types.
The SIHF and the SIPFs for the dwelling should possibly be combined, as they are in the 'mainstream' form. However, the summary information gathered on 'absent usual residents' and 'visitors' is valuable, and should be retained. The SIHF might be retained as a separate document, but in another guise, as a 'household checklist'. This is how it was used by the enumerators at community A, and it proved a useful tool.
The layout of the SIPF needs attention. The use of large, bold print versus small print should be carefully reviewed, bearing in mind that many Indigenous people (including many enumerators) are not fluent readers of English. The repetitive demands to supply postal addresses (for dwellings and places of work) should be circumvented.
The design of the pathways through the form also needs some further thought. One possibility is to uncouple CDEP from the labour force data so that CDEP participants do not have to answer numerous questions about their 'work'. The questions on further education and training could also be streamlined.
More thought should be given to devising questions that allow 'non-mainstream' aspects of Indigenous social life to emerge from the data. In some cases this simply involves putting more thought into the prompts to a question, to allow a wider range of options in the response. The overall impression given by the data on the SIPF is that the people of community A 'lack' certain things in comparison with the mainstream—education and 'real jobs' in particular. If the concept of Australia as a multicultural society is to be taken seriously, then an instrument such as the census should be attempting to capture, as far as possible, significant features of non-mainstream cultures that the mainstream culture lacks, particularly those that influence interaction with the mainstream in significant domains such as welfare, employment and education.
Further to this, a certain amount of 'deconstruction' of mainstream categories would not go amiss. The 'nuclear family' is a case in point. Another is the way in which different sectors of the population are categorised in the questions on 'origin' and 'ancestry'.
Many of these suggestions, if taken up, would result in a shorter and more 'culture-specific' form, tailored to those Indigenous people who live in discrete communities on or near their traditional 'country', with limited or no access to the labour market. These are communities where the major organising principle of social life is kinship, where households tend to be large, compositionally complex, multi-generational and somewhat fluid in composition because of high levels of individual mobility, and where non-mainstream forms of 'work' make an important contribution to the local economy and to cultural life. This might, perforce, narrow the scope for the administration of the IES by enumerators. Other- 73 - Indigenous people might be candidates for a self-administered IES form, or would be required to fill in the mainstream form, with help from an enumerator if needed.
Such a strategy would not be without its own problems: Indigenous Australia was never homogeneous, and the contemporary situation there is even more diversity as a result of the vagaries of the colonial process and the uneven penetration of the settler culture and its institutions. The terms 'remote', 'rural' and 'urban', with which the settler society attempts model this diversity, once reflected a certain logic of distance. Today, however, these terms are becoming increasingly metaphorical. A 'remote' community may find itself cheek by jowl with the company town—and the mine—of a multinational enterprise. 'Rural' Kuranda and 'remote' Yuendumu are different in many respects, but have a 'number of fundamental commonalities', including the fact that: 'the concept of "family" based on the extended family formation is the central and abiding social and economic construct, and a key component of individual identity' (Smith 2000b: 95). The ABS would have to assess each region, each community within a region, and even individuals within communities to determine which was the most appropriate means of enumeration for them. But the result would be a far higher proportion of self-administered forms, and the collection of much more reliable and relevant data on a sector of the Australian population that will continue to be not only highly distinctive but also in great need of well informed state policy and programs.