Table of Contents
I made several visits between November 2006 and March 2007 to the Data Processing Centre (DPC) in Melbourne, to observe the work of the Indigenous Processing Team (IPT). The creation of the IPT was an innovation for the 2006 Census, with a cohort of data-coders trained to deal specifically with the Interviewer Household Form (IHF). All Collection Districts (CDs) of Types 11, 12 and 13—all those consisting of or containing discrete Indigenous communities—went through the IPT. Since CDs of Types 11 and 13 also contained other kinds of communities that were enumerated via the mainstream form, the IPT coders had to deal with both kinds of forms.
I observed aspects of most stages of the data processing, from the grooming of the forms and the compilation of the Census Record Books (CRBs) before the electronic capturing of the data to coding of the first and second-release data.[1] I also attended the IPT coders’ training sessions on the processing of household and family data, and of data relating to occupations and qualifications. I was given the opportunity to meet with individuals responsible for instituting and overseeing general systems and procedures within the DPC, and this helped me to gain an overview of the Indigenous Enumeration Strategy (IES) coding within the broader context of DPC activities.
As a field site, the DPC was very different from the remote area where I observed the enumeration and from the Darwin Census Management Unit (CMU). During the enumeration, I had been working in an environment that I knew well and where I had undertaken a similar exercise in 2001—and, as a result, I had an overview of the process that was informed by my prior local knowledge. Also, since the filling in of the IHF was a protracted exercise, it was possible to make very detailed observations of the initial data collection, in ‘real time’ as it were. At the CMU, although the institutional environment was initially unfamiliar, I was observing a relatively small-scale operation—in contrast with the DPC—and one in which, although use of information technology was a significant component, face-to-face interactions—for example, in the training sessions—and manual processing of the IHFs (with people talking about what they were doing as they worked) were important elements of the process. The data were also arriving at a pace and in a form in which I could ‘capture’ it for the purposes of my own analysis.
In contrast, the DPC is a very large and complex—and highly technologised—environment. I had to rely largely on others for the information that would allow me to understand its workings, rather than observing for myself. Although I was able to observe the training of the data-coders in a manner similar to my observation of the training of the Census Field Officers (CFOs), once they started work the situation was very different. Whereas in the field and at the CMU the data were always in context—physically on a form that was in a box with other forms from the same place—once the data from the forms had been captured electronically at the DPC it was dissociated from those contexts. Although the coders could call up the electronic copy of the form if need be, for the most part they were working, at high speed, on snippets of information divorced from the wider context of the forms. In order to meet processing deadlines, they had to concentrate hard and work fast. There were fewer contexts for me to observe people’s own commentary on what they were doing, and—with one exception, which I will note below—the data were not ‘capturable’ in the same way as in the earlier contexts.
In my work at the DPC, I was concerned less with analysing the organisational aspects of the exercise than was the case in the field, although I will make some brief general comments based on my rather superficial knowledge of what was a very complex and technologically sophisticated operation. My primary aim was to follow the progress of my own case-study IHFs, so I would have a complete picture of the journey of the data through various contextual frames, from its elicitation during the count to the coded end product.
There was not much more to observe as far as the basic ‘head count’ aspect of the census was concerned. The IES team did run another check on the internal consistency of the records while compiling the CRBs during ‘pre-capture’. They also checked whether people who had been moved from ‘persons temporarily absent’ (PTA) status to ‘inside the form’ had then been eliminated from the PTA list at Question 11. By and large, this was not a problem with the Northern Territory forms—thanks to the work of the Darwin CMU—but quite a few ‘duplicate’ people were found in the IHFs from other States. In such cases, the person was usually retained inside the form at Question 12 and eliminated from the PTA list unless it was clear that they would have been counted elsewhere, for example, at boarding school. There were a few cases where the sex of individuals was missing, and these were imputed.[2]
I was most interested in the coding of the data that had as yet received little attention at any stage of the checking process in the field or at the CMU: this was the primarily socio-demographic information—household and family composition, language use, education, employment and so on. Many of the questions on the form relating to these issues required a written answer and, for the IHF, answers to such questions were coded manually. This was a major difference between the coding procedures for the standard form and the IHF.[3]
I was interested in the categorisations that underlie the coding process and how the data were fitted to those categories. As I have argued elsewhere (Morphy 2007), the categories in terms of which the national census is framed are derived from the culture of the mainstream and reflect the concerns of the nation-state. In Villaveces-Izquierdo’s (2004: 178) words: ‘[T]he census is a tool through which the state envisions and acts upon the nation.’ These categorisations, which also underlie the framing of the questions on the form, are opaque to those who fill in the forms—the collector-interviewers (CIs) and the interviewees.[4] In many cases therefore, the answers given to questions—particularly where written answers are required rather than simply the ticking of a box—are often difficult to interpret in terms of the preset coding categories. This was the main reason why such responses were coded manually.
In the field, I had been interested in the categories that Yolngu brought to bear in responding to the census questions, and what kinds of answers this produced. At the DPC, I was interested in how their answers were interpreted and slotted into the coding categories. I was interested in the demographic portrait of the Yolngu population that was produced as the end result of this process, and the degree to which it was commensurable with: a) the Yolngu view of themselves, and b) a depiction informed by anthropologically derived categories.
Some might argue that this is an unnecessary and even misconceived exercise, that the purpose of the census is, precisely, to gather demographic ‘facts’ that are comparable between different sectors of the population. My argument will be that, to the degree that these ‘facts’ are socio-demographic rather than socially neutral, they will be categorised differently depending on the cultural lens through which they are viewed. Census categories are not culturally neutral, and it cannot be assumed a priori that the categories of one socio-cultural system are translatable directly into those of another. To some extent, then, this chapter is less a commentary on my observations of the operation of the DPC and a more broad-ranging commentary on the nature of census data and its limitations.