Clearly, enumeration of any kind in remote areas is a difficult task. Compounding the problems presented by dispersed settlement over large distances is the high mobility of the Aboriginal population in such areas. Furthermore, the geographical and social distribution of households and their individual members present definitional problems (Henry & Daly 2001; Smith 2000a). The obstacles presented to accurate enumeration by these and other factors help to explain the gap that is sometimes observed between official and unofficial population counts, and between the statistical construction of social forms and the reality on the ground. Such conundrums would be resolved to some extent if documentation were available detailing the method and practice of enumeration. At the very least, those using data would be more aware of how they were acquired, thereby allowing greater precision in their interpretation. There are several unique features of the remote IES that have potential bearing on the quality of census output and of which users need to be appraised. These provided a focus of attention for the field observations of CAEPR researchers.

For pragmatic reasons—the fact of high intra-regional and intra-community mobility—two of the most fundamental features of standard census procedure in Australia are effectively overridden by the remote area strategy. In the standard enumeration as applied to the general population, a self-administered count occurs on a single day—census day (7 August in 2001). In contrast, while referring to the same census day, the process of remote area counting can begin weeks before the day itself and continue for some time after.

The second departure from normal enumeration procedures concerns the nature of the count. The Australian census is, in the first instance, a place of enumeration, or de facto count. There is some indication, however, that the Northern Territory office of the ABS encourages a place of usual residence, or de jure count. At the same time it seems likely that a mix of de jure and de facto counting may occur in all jurisdictions since individuals are recorded in a manner often dictated by expediency. A complicating factor is the question of who constitutes a visitor to a household and whether CFOs and CCs make assessments on this matter in a sufficiently coordinated way.

Another departure from standard census practice is the use of interviewers and specially designed forms and questions. While it was not possible for each researcher to fully observe the training of census collectors, observation of their interactions with respondents revealed much about its effectiveness. Direct observation also provided insight into respondent attitudes to census-taking as well as a host of related issues. These include problems posed in acquiring the personal details of household members in often very public situations, misunderstandings about the nature and meaning of certain questions, the struggle to translate Aboriginal conceptions of the social and economic world into those demanded by the questionnaire, and interviewing difficulties presented by the ordering of some questions and the length of time taken to interview. In observing these - 11 - difficulties, particular attention was paid to any trade-offs that might occur between an ability to achieve the fundamental purpose of the census, that of securing accurate basic demographic information, and the increasingly burdensome task of detailing social and economic characteristics.

The costs and benefits of devolving control of field procedures and decision-making to local units was also under scrutiny. Census-taking in remote communities presents a logistical task of considerable proportions, and very often this demands approaches that can best be described as innovative, or at least individualistic. In observing the playing out of these strategies, it was possible to assess the degree to which flexibility in an otherwise highly structured system either enables or enhances enumeration. Also highlighted are the sorts of obstacles that collectors face. One issue that these observations illuminate is whether consistency applies across the count, either as a de facto or as a de jure enumeration. The intent of the ABS, centrally, is that it should be the former.

The three independent case studies that follow collectively shed light on the issues outlined above. In each case, a common methodology was applied involving (where possible) observation of training sessions for ABS staff and interviewers, 'over the shoulder' observation of actual census interviews, structured interviews with CFOs, CCs, and census interviewers (and officials of community councils where appropriate), observation of interactions between interviewers, CFOs and CCs, and observation of the organisation and administration of census logistics. While styles of inquiry and reporting vary, reflecting different disciplinary backgrounds (Martin and Morphy in anthropology, and Sanders in political science), the studies complement one another to present a holistic picture. It is interesting to note that a high degree of similarity is observed across quite separate geographic settings both in the range of practical issues faced by census collectors in pursuing their task, and in the responses of Aboriginal people to the exercise. These common findings, together with local nuances, are summarised in a final chapter with a view to considering future options for improving the census in remote Indigenous communities.