1. Producing powerful numbers

In 2001 a team of CAEPR researchers observed the conduct of the census in three remote Aboriginal communities (Martin et al. 2002). While we were supportive of the interview and time extension adaptations of the 2001 IES, we observed that the two-form structure of a household form and separate individual forms then in use was very cumbersome and made large, somewhat unnecessary administrative demands on the local recruited census field staff—the Community Coordinators (CCs) and the collector-interviewers (CIs). In two of our observed 2001 locations—one in the Northern Territory and one in Queensland—the two-form structure was made to work largely by keeping household forms in the background and exposing interviewees only to the personal forms. In the third location observed in 2001, 12 days of very slow progress led to abandonment of the personal forms and a salvage operation that focused simply on the household forms. The two-form structure was just far too demanding on the interest and persistence of the interviewers, and their interviewees. CIs were becoming burnt out and falling by the wayside in the process of enumerating just a limited number of households.

Frances Morphy’s work in 2001 (Morphy 2002) focused in addition on the highly inadequate construction of Indigenous households through particular census questions. She judged that the attempt to translate Indigenous kinship systems into Western terminology had been largely unsuccessful and that many of the Indigenous household descriptions in the census data were, as a consequence, of limited value. Morphy also focused on the ‘legendary mobility’ of Indigenous people in her observation area in north-east Arnhem Land and how, together with time extension of the collection process, this meant that a ‘complete enumeration’ was almost impossible. Double-counting and under-enumeration, she reasoned, were highly likely, but the extent of each would be difficult to assess.

An awareness of these Indigenous mobility issues had led the Northern Territory administration of the ABS, over some years, to attempt to count ‘usual residents’ of Indigenous communities, rather than following the standard census procedure of counting people present, with some provision for adding absent usual residents who might not be counted elsewhere. Morphy sympathetically considered the merits of counting usual residents, but ultimately argued against it on the grounds that a robust definition of ‘usual resident’ was hard to develop. Will Sanders’ work on the Alice Springs town camps argued against the usual residents approach because it allowed people present—labelled ‘visitors’—to slip through the enumeration process.

One final common theme in our 2001 observations was the limited social relevance to the circumstances of traditionally oriented Aboriginal people in remote Australia of many census questions and their pre-specified categorical answers—whether this was the difficulty of defining a dwelling and its associated household, of categorising an education level or course undertaken, or a question about religion or marriage, employment or looking for a job. Often this limited social relevance of questions and categorical answers would introduce humour to the collection process—for interviewers and interviewees. Lack of social relevance could, however, also lead to disinterest and disengagement.

On the basis of all these observations, in the final chapter of our monograph on the 2001 Census (Martin et al. 2002), we argued for essentially three reforms to the IES as we observed it operating in remote Australia. In opposition to some half-hearted moves in the Northern Territory towards enumerating usual residents, our first suggestion was to argue for a return to the general ABS approach of counting people present, with some facility for adding absent usual residents who might not be counted elsewhere. Our second suggestion was to argue for the reintegration of the special Indigenous personal and household forms into a single Indigenous household form designed to be administered by interview. Our third suggestion was for this form to be tailored more precisely to the circumstances of Indigenous people in remote Australia by the restriction of some questions asked and the development of more appropriate categorical answers.

The ABS formed an IES Working Group after the 2001 Census to consider ways of improving field design and methods. Sanders and Morphy were invited to be members of this group. It produced two key initiatives—one conceptual, the other practical—which corresponded with our first two ideas for reform. First, in 2006 there was to be a clear move back to the standard of enumerating people present in a dwelling at the time of the count, plus absent usual residents judged unlikely to be counted elsewhere.^[1] Second, the previous multi-form schedule of census questions was to be integrated into a single matrix-style Interviewer Household Form (IHF). Our third idea for reform, however—attempting to restrict this new form to a lesser number of questions of greatest social relevance to the circumstances of Indigenous people in remote areas—proved almost impossible for the ABS to implement. While individual questions and categorical responses to them could be modified slightly, the idea of leaving out any questions always met resistance. Any question left out would become a statistic for which there would no longer be a national Indigenous/non-Indigenous comparison. This was understandably very hard for the ABS to contemplate.

A primary purpose of observations in 2006 was to test the workability and impact of the reforms discussed above. In addition, the opportunity was presented for the first time to assess the nature and effectiveness of pre-census preparations and post-census processing. Two of the locality studies revisit places observed in 2001, while two others cover places being observed for the first time. The monograph also contains three chapters that observe related administrative processes leading up to and following on from enumeration in localities; preparing and undertaking training for the enumeration in the Darwin Census Management Unit (CMU), checking forms in the Darwin CMU after the field enumeration and, finally, the processing of these forms at the Data Processing Centre (DPC) in Melbourne (see also Morphy 2007). Our study this time has therefore been widened to include observation of ‘back-office’ administrative processes before and after the event, as well as enumeration in the field.

During the course of the research our focus shifted—as a result of what we were observing—to the broader structures and processes of the IES. The question of who was being counted where loomed ever larger as we watched the struggles of the Census Field Officers (CFOs) and their inadequate numbers of recruited field staff to maintain control of the process, and an orderly count, in the context of a prolonged engagement with a highly mobile population. We came to see that the strategy as presently conceived is ill-equipped to deal with the agency of a population for whom the census is essentially an unfathomable state project. It does not engage adequately with the resources and local knowledge embodied by the organisations that straddle the interface between these populations and the state. Nor is it designed to cope realistically with the contingencies that arise in everyday life in these remote communities.