Let me conclude by pointing to some features of the Indigenous statistical archive in its reformed, contemporary condition.

First, there has been a sustained agenda of reform. This agenda has been driven by a strong conviction that social justice demands Indigenous/non-Indigenous comparison across many socioeconomic and health variables. It has been notably effective in the reform of State and Territory registrations of vital events. Len Smith says that agitation on this point began around 1965 (Smith 1982: 16). According to the ABS (2000a: Table 11.6, p.162), these are the periods within which the various Australian governments put an Indigenous identifier into databases relevant to demography and to population health analysis:

On at least one occasion, the reform of administrative data was stimulated by the demands of a focused inquiry with statutory powers. When the Royal Commission into Aboriginal Deaths in Custody (RCADC) began, its staff knew they could draw on the prison censuses that had begun to use an Indigenous identifier in 1982. But what data was there about police custody? The Royal Commission used its prestige and legal powers to persuade all of Australia’s Commissioners of Police to allow the research staff to conduct a National Police Custody Survey in August 1988 (Commonwealth of Australia 1991: 191–2). Analysis of these data led to a conclusion that determined much of the Commission’s subsequent agenda: that deaths in custody were disproportionately Aboriginal not only because of factors within custody but also because Aborigines were much more likely than non-Aborigines to be in custody. What accounted for this higher rate of incarceration, the Commission asked? The Commission postulated the related concepts of ‘self-esteem’ and ‘underlying issues’ in order to adduce research on education, labour market status, health and other factors (using data from the 1971 to 1986 censuses) to explain Aborigines’ disproportionate entry into police or prison custody.

Second, the reformed statistical archive makes some use of the organisational capacities known as ‘the Indigenous sector’, for example in Community Housing and Infrastructure Needs Surveys (CHINS) of 1992 and 1999 and in a current Mental Health Survey. The 1999 CHINS collected data from 707 Indigenous housing organisations and 1291 ‘discrete communities’ (1089 of which had an identified housing organisation). ‘Data … were collected through personal interviews with key members of Indigenous housing organisations and communities who were knowledgeable about housing and infrastructure issues. Such people included community council chairpersons, administrators, coordinators, clerks, housing officers, water and essential service officers. Information regarding health services was generally collected from health clinic administrators.’ (ABS 2000b: 62).

Third, there is a continuing concern to improve the methodology of data collection in remote and very remote Australian communities (Martin et al. 2004).

Fourth, we risk becoming data rich and theory poor. Our data are useful only if we have some theoretical framework in which to make sense of them. Perhaps the most developed theoretical model that we have, at the moment, is that which I mentioned in my introduction. The Productivity Commission’s 2003 Report operationalises the ‘strategic areas for action’ in terms of thirty variables on which Indigenous and non-Indigenous can be compared (see Steering Committee for the Review of Government Service Provision [SCRGSP] 2003). The theoretical model at the centre of this strategy highlights the familial and educational conditions of Indigenous childhood that are conducive to (or destructive of) an Indigenous person’s later fortunes in the labour market. It would be a useful exercise to compare this theoretical model with predecessors such as Rowley (1966) who drew on North American studies of the pathologies of institutions, and the Royal Commission’s, with its focus on the material determinants of low self-esteem. That is the task for another paper. One of the questions that such a paper might address is how ‘Indigenous culture’ is defined as a quantifiable variable in our (more or less conscious) theoretical models, and what significance, if any, is attributed to it.

Fifth, we are producing more and more data at a time when the government is terminating the existence of at least one of the bodies—the Aboriginal and Torres Strait Islander Commission (ATSIC)—that should be the primary users of the data. I endorse the point, made by a number of authors in this volume, that one of the best reasons for having a rich Indigenous statistical archive is to enable the Indigenous sector to talk back to government about policy and about program effectiveness.

Sixth, in the often angry, frustrated and demoralised public discussion of Indigenous affairs policy, we now have two kinds of language for talking about Indigenous disadvantage. Here I would like to invoke the work of the Canadian philosopher Charles Taylor. In his recent book, Modern Social Imaginaries (Taylor 2004a), he says that we moderns have developed the ability to think about social life in two distinct ways, both of which are valid and useful. On the one hand, we can think about society as a series of interactions between variables. There are several papers in this monograph in which that was the idiom for thinking about society. The Productivity Commission Report is, again, a wonderful example of that way of thinking. On the other hand, we can think about society as an interaction between responsible, intentional agents, such as individual people, and organised collectives of people or organisations, including governments. Taylor says that the modern social imaginary—that is, our taken-for-granted ways of thinking about ‘society’—is ‘bifocal’; we use both mechanistic and agent-centred thinking. We imagine society in mechanistic terms—interactions between variables—and we imagine society humanistically—interactions between thinking, feeling agents who can be held responsible for what they do.

In contemporary Australia, both ways of thinking about society have been recently intensified. On the one hand, due to the reforms of the statistical archive, we have data with which to think about the interactions among an increasing number of variables. On the other hand, we have the language of welfare reform, with its emphasis on ‘mutual responsibility’ and ‘Shared Responsibility Agreements’ (SRAs). This language also has a horror of any ‘welfare’ that is ‘passive’ and a strong implication that Indigenous Australians may fail in their responsibilities to take up the opportunities that the public and private sector provide. In contemporary Australian public culture, the mechanistic social imaginary that is fortified by our rich statistical archive is in daily juxtaposition with the voluntaristic social imaginary that asks: are governments living up to their responsibilities to Indigenous Australians and are Indigenous Australians taking responsibility for their own advancement?