Whereas most, if not all, non-Indigenous Australians know the precise date and year of their birth, even if they cannot remember their closest relatives' telephone numbers, exactly the reverse is true of most local Indigenous people. It is a question of what it is considered important to remember. The enumerators made valiant efforts to ascertain people's ages, and E1 was intending to check the answers later against the clinic records. I do not know if this actually happened, but if it did, it would still not have resulted in accurate date-of-birth information from some people, particularly the older ones. For example, one man gave his age (emphatically) as 63. His driver's licence records his birth date as 01/07/45, and the HA clinic records have his birth date as 01/01/48. Many older people have the first of January (or the first of July) as their birthday in the clinic records and on official documents. It can safely be inferred that such 'dates of birth' are guesstimates made by missionaries and other local record keepers.

There are local terms which designate degrees of maturity, or stages of life: [baby], [child], [circumcised boy], [prepubescent girl], [young man], and so on. Some terms from the mainstream have also been adopted locally, for example 'preschooler', 'schoolkid', and 'pensioner'. These terms all relate indirectly to age ranges. It might be better, therefore, to ask this question in a different way, giving people a choice of boxes to mark according to whether they were 0–5, 6–10, 11–15, and so on. It would be possible then, in training, to get the enumerators to match these up with the local terms for particular life stages. These could then be used to elicit the age data from the interviewees.^[20]

Because the census is focused on the dwelling as the unit of analysis it cannot capture relationships between people who live in different dwellings. Does this matter? Every analyst who uses census data is aware that it is, above all, a dwelling-based survey and that it does not seek to provide data on inter-dwelling relationships.

However, many analysts may not be aware that Indigenous households can and sometimes do have members in more than one dwelling. In community A, as in many other Indigenous communities, the households of certain dwellings are linked closely to one another—economically as well as by ties of kinship. This is particularly true of dwellings containing a man and dwellings of the wives with whom he is not presently cohabiting. The people at community A rejected 'separated but not divorced' as the correct box to mark for this scenario. The SIPF is neutral on the question of polygamy (either by design or by accident). It is not possible to answer the question on marriage (Q. 6) in a way that indicates how many wives a man has, or how many co-wives a woman has. As we have seen, at- 68 - community A there are, apparently, two dwellings where person 1 is a married woman with children, but where there is no 'husband' on the scene. These data might be interpreted to indicate the presence of 'single parent' households, with all that this implies in mainstream society. This would be a misinterpretation, since these are not deserted and disadvantaged women struggling to raise their children in isolation. They and their children are part of an extended family that happens to occupy more than one dwelling—and indeed it is not uncommon to find some of the children of one co-wife living in the dwelling of another.

The dwelling is not the salient 'unit of residence' at community A; rather the community itself is. This much was clear in the enumerators' and respondents' reactions to Q. 7–9, with the recurring demand for postal addresses. Even larger communities such as B are almost invariably divided into recognised sub-communities or 'camps' (e.g. Top Camp, Bottom Camp). The design of the SIPF should perhaps acknowledge this fact, by using 'community' or 'sub-community' instead of 'place' and by eliminating requests for postal addresses. The subdivisions of big communities could be identified as part of the training of the enumerators.

The data collected at community A on where people were living one and five years ago are almost certainly unreliable. One problem is the concept of 'X number of years ago', since local people do not measure their lives in such terms. Solutions to this problem might be found in the training of the enumerators, as in the case of the 'age' and 'place' questions. It might simply be a matter of impressing upon them that the ABS considers this information to be important, and devising more appropriate, locally salient cues, such as particular memorable events that were of importance to community members.^[21]

There is another, more intractable problem: for local people, these questions about 'place' are not value-free, as they are for most non-Indigenous Australians. Place is not just physical space. People have strong spiritual and emotional attachments to particular places, and if they are now living at one of those places they are more than likely to state that this is where they have always lived 'most of the time', thus understating their mobility.

The questions on origins and ancestry need rethinking. In their present form they are underpinned by covert distinctions—'Indigenous' as opposed to 'ethnic' as opposed to 'unmarked' (Anglo-Celtic mainstream). Of these, the 'Indigenous' is the most marked, because it is referred to in two separate questions, and the Anglo-Celtic mainstream is so 'unmarked' that it is not even mentioned. From the Indigenous perspective this makes no sense, and although the responses were accurate—or appeared to be—the questions were felt to be decidedly odd. For Indigenous people, 'Indigenous' is the unmarked category, and the Anglo-Celtic mainstream is the most marked, because it is the most common source of non-Indigenous ancestry for local Indigenous people.

What precisely is the information being sought by the question on origin? It is not clear to me, and it certainly was not clear to the local Indigenous people. Is it another question about ancestry in a different guise, or a question about self-identification?

It is important for the designers of the IES to be aware of issues that are potentially loaded in a political sense. The bilingual education program in the Northern Territory was recently under real threat. One argument that can be used to challenge the viability of such programs is that it is just too difficult to cater for communities where many different languages are spoken. This is not an argument that can seriously be mounted at communities where the language in question is spoken, but the data from the 2001 Census, in the hands of non-specialists, could be used to support this argument. It would have been a simple matter to instruct the enumerators to enter the term that has been coined for the language (by linguists) as the language spoken by all speakers of a dialect of that language. In terms of providing salient information for policy-makers this would have been the best option (although for the sociolinguist the data that were actually collected are much more fascinating).

The 2001 data issued by the ABS in its first release indicates that nearly 20 per cent of Indigenous respondents at HA outstations said they spoke 'English only'. This does not correspond in any way to reality: I can think of no local Indigenous individual in this region who does not speak an Aboriginal language as their first language.

If the question on religion had carried the explicit instruction that more than one box could be marked, the answers would have been a more accurate reflection of the situation at community A. The term 'Traditional Beliefs' carries an implicit value judgment: beliefs do not have the same status as religion unless they are regarded (and described) as a 'belief system'.

The data in the first ABS release of the 2001 Census indicate that fewer than 40 per cent of the Indigenous population of the HA outstations marked 'Traditional Beliefs' as their religion. This does not accord with my own observations of the Indigenous people of this region, who continue strongly to maintain their traditional belief system and their ceremonial life. The design of the question on the SIPF almost certainly materially influenced the response.

Most of the questions on schooling and further training and education were difficult for community A enumerators and residents to understand and answer, and this generalisation probably holds for many Indigenous communities in 'remote' Australia. Yet arguably, accurate and interpretable data are vital for policy formulation in this area, which will have a strong influence on the future wellbeing and self-sufficiency of discrete Indigenous communities. No matter how the questions are formulated, the training of the enumerators needs to focus on their content. The enumerators need to be given the knowledge to interpret the responses of the interviewees, most of whom have not received formal education past the primary level of schooling. Those who have further training of some sort mostly fall into the category 'Yes, other course' in Q. 23. In such cases it should be possible to sidestep the next four questions.

Most adults at community A are on CDEP. The status of the CDEP as a provider of 'real work' is politically contested (see e.g. Shergold 2001). The SIPF for the 2001 Census unequivocally equates CDEP with other kinds of work, starting at Q. 29 with the question: 'Did you have a paid job last week?' and providing a box for 'Yes, worked—CDEP'. In the questions which follow, almost to the end of the form (Q. 30–38), the interviewee is led, inexorably, further down this path. The SIPF makes valiant attempts to keep all these questions relevant to CDEP 'jobs' but the effort shows, for example in Q. 34: 'What work does your employer do? (If worked for CDEP write 'community council')'.

The way in which these questions are put creates a double bind for local people, because CDEP is not primarily payment for work in the outstation context—at least not for the kind of work contained in the prompts to Q. 30. CDEP payments are nevertheless vital to people who live on outstations, since they are the main source of regular cash income. It is therefore quite unsurprising that the answers to the questions on 'work' were formulaic.

Although the 'rubbish collectors' certainly keep this particular community clean and tidy, the job does not represent 20 hours of work per week for several young men. The inference that may be drawn by some is that CDEP is 'sit-down' money. The inhabitants of community A do not spend an undue amount of time sitting around doing nothing, but the prompts for Q. 30 create no space for people to mention the work that they spend most time doing. A point made by Diane Smith is pertinent here:

Hunting and gathering are daily activities for most young 'rubbish collectors' at community A. And if these are implicitly discounted as work, how much more so is working for the community during ceremonial performance? The same young 'rubbish collectors' spend sometimes up to six hours a day dancing energetically and skilfully in the hot sun, for days at a time. Such skills, and prowess at hunting and fishing, are the product of long hours of learning and practice, and the work that results contributes significantly to the physical and cultural wellbeing of the community. The focus in the SIPF on the mainstream notion of 'work' devalues these 'traditional' forms of work. The failure to provide a space for their inclusion does two things: it implies that this kind of work does not contribute to the economic and social wellbeing of the individuals and their community, and it allows those who are so minded to characterise Aboriginal people as getting 'something for nothing'.

There are two possible solutions. If CDEP is categorised as work, then traditional forms of work, such as hunting, fishing, collecting bush tucker and making art should be included in the prompt to the question: 'What job did you do?' And it should be possible to list more that one job.

Alternatively, CDEP in places like community A should perhaps be uncoupled from the notion of work in the mainstream sense, and from the labour force statistics. A desirable side effect would be that the SIPF, for people on CDEP, would be much shorter. If this solution were adopted there should be an additional question on the SIPF that allows people to estimate the extent and nature of their 'unpaid' work. Art and craft production, which is one of the only other significant sources of cash income, should be included in the prompts for non-CDEP work: its presence might cause (a few) people to reflect a bit more closely on the sources of their cash income, and to make some effort to record them.