One major issue for analysis relates to how much the special enumeration strategy for the census demands of Aboriginal people as both interviewers and interviewees. The Alice Springs town camp experience in 2001 suggests that the strategy demands far too much of both interviewers and interviewees. The SIHF and SIPF structure is cumbersome and time consuming, and greatly extends the time that both interviewers and interviewees have to dedicate to the task. There is a constant danger that interviewers and interviewees will lose interest in the process before the enumeration is complete. People who have seen others being interviewed, or have other things to attend to, may not offer themselves for interview, while interviewers may burn out after doing just a few households.

To maintain the interest of both interviewers and interviewees the ABS probably needs to develop a simple single-form interview procedure which can be administered quickly. This single form structure could either be an enhanced household form, not unlike the one devised informally half way through the 2001 Alice Springs town camp enumeration process, or a simple personal form, not unlike the special short form used in the homeless enumeration (see Appendix D). The choice between these two single form structures would depend, to some extent, on whether the ABS sees the census as primarily concerned with enumerating and identifying the characteristics of individuals or of households. This is, of course, not entirely an either/or choice. A personal form could ask where people lived, how many other people lived there and whether the interviewee was related to those others, while a household form will inevitably collect some pieces of personal information. But it seems, on the basis of the Alice Springs town camp experience in 2001 that a personal form should be designed to be completed in no more than five minutes and a household- 88 - form, for perhaps ten people, in no more than half to three-quarters of an hour. Otherwise momentum and interest—and people—will be lost.

Under the current two-form structure, by contrast, the enumeration of a household of ten can take up to two and a half hours, with people identified on the SIHF at the beginning expected to hang around for later completion of a SIPF. This is simply too much to ask and the count suffers as a result, through both interviewer burn out and interviewee avoidance.

Considerable thought would need to be given as to which single form structure, and associated personal or household collection emphasis, should be applied to discrete Aboriginal communities in sparsely settled northern and central Australia.

In the course of undertaking these observations, it was often suggested to me by the central Australian census manager that census collection in the Alice Springs town camps could, in many ways, be seen as something of a worst-case scenario. There is clearly a very high degree of mobility between the camps and the outlying Aboriginal communities. A good deal of drinking goes on in the camps, which makes collection difficult and at times unpleasant. The daily mobility of town camp residents out of the camps into other areas of Alice Springs is also a problem, making people hard to catch in the dwellings. The spread of the camps across town also makes travel to and from them, or between them, difficult for interviewers.

Despite all these worst-case characteristics, it seems to me that there are also aspects of the Alice Springs town camp situation that are very positive and helpful. Tangentyere is a well-established, cooperative, and able Indigenous organisation which cooperated well with the ABS. The interviewers recruited through Tangentyere's Job Shop and its general networks were also of high quality, being people who had through their past employment and other experience considerable exposure to non-Indigenous bureaucratic ways. On these latter counts, the situation in some remote Indigenous communities could be more challenging than in the Alice Springs town camps, with less well-established community organisations and less bureaucratically-experienced individuals on whom to draw as interviewers.

One further characteristic of the IES as implemented in the Alice Springs town camps was its emphasis on enumerating people who are usual residents of dwellings, plus visitors who were unlikely to be enumerated elsewhere. This is somewhat different from standard census practice, which is to enumerate people present on the night the census, plus absent usual residents who are unlikely to be enumerated elsewhere.

Formally, the IES strategy followed the standard census approach. This was reflected in the SIHF, which asked interviewers to list all people who live in the dwelling 'most of the time' including those who are 'away' and also all visitors. The SIHF then asked the interviewers to divide these people into those who needed a SIPF (i.e. to be enumerated here) and those who did not. And the accompanying training manual explained as follows:

- 89 -

The emphasis here was on identifying absent usual residents who would be counted elsewhere and hence did not need to be counted here, to avoid double counting. But all people present, including visitors, were to be counted on SIPFs.

Informally, the Northern Territory census administration had tried to move more to a usual residents basis of enumeration, counting all usual residents including those absent. Visitors were only to be counted if it seemed likely that they would not be counted elsewhere as usual residents. In practice this seemed to mean that interviewers let visitors go, either not recording them at all, even on the SIHFs, or not filling in SIPFs for them.

This seemed to me a highly unsatisfactory resolution of the issue of who should be counted. People present were not being counted on the premise that they would be counted elsewhere as absent usual residents. However, this assumption seemed suspect on a number of counts. First, would they still be recognised as usual residents of a dwelling elsewhere, perhaps out in a bush community, if they had not been there for a while? Second, if they were recognised as absent usual residents somewhere else, would a SIPF be filled in for them? I had observed some willingness to put people not present on the list of household members for the SIHF, but a reluctance to fill in a SIPF for other than oneself. The personalised 'you' language of the SIPF seemed to encourage this reluctance. Third, and specifically in relation to the town camps, it was arguably just as important to count the population present as it was to count those who identified themselves as usual residents.

I would suggest that the Northern Territory census administration's attempt to move to a 'usual residents' basis for enumeration was somewhat half-baked and that in the Alice Springs town camps it probably allowed a lot of visitors, who might not in fact have been counted elsewhere, to slip through the census net. There seem to me to be very good arguments for sticking with the standard census procedure of enumerating people where census collectors encounter them, plus asking those present if there are other absent usual residents. Those present can be asked whether they are usual residents or visitors and if the latter, where their usual residence is. Those absent can be reasonably clearly divided into those who are likely to be enumerated elsewhere and those who are not. But visitors are extremely difficult to divide into those who are likely to be enumerated elsewhere and those who are not. To let the visitors go when they are encountered, on the assumption that others elsewhere will be asked about them and will answer for them, seems perverse. Surely the best source of information about a person is that person.

The attempt by the Northern Territory census administration to move to a usual residents basis of enumeration for discrete Aboriginal communities is, I think, linked to the underlying issue of the time extension of census collection under the IES.

- 90 -

Standard census enumeration aims for a very short time extension to the count, enumerating people where they sleep on a single night. This can only really be accomplished if most of the count is done by self-enumeration. Where the count is done by interview, the time extension is inevitably going to be greater. Time extension, combined with mobility of the population, introduces a new source of possible counting error: missing people or encountering them more than once as they and the census collectors move around over time. Hence the drive towards identifying usual residents and discounting visitors. However, the significance of error due to time extension and mobility can be overstated. Error is reduced if one counts those present at the time of enumeration and asks, among other things, whether they have been counted elsewhere recently. There may still be under-enumeration due to missing some people altogether through mobility, but questions about absent usual residents can pick up on this too. Time extension does not need to entail a switch to a usual residents basis of enumeration. Enumeration can still be of those present, plus absent usual residents. The question that needs to be asked in a time extended census is not who slept in this dwelling on the night of 7 August or some other specific date, but who slept in this dwelling last night. In this way in a time extended census all dwellings get a census night, even if it is not quite the same census night as the dwelling next door or that in another community.

In relation to individual questions, like those which caused concern for collectors and CCs in the initial training course, the following comments can be made. These all reflect, in one way or another, on what might be referred to as the social relevance of these questions to people living in discrete Aboriginal communities in sparsely settled northern and central Australia.

The attempt to get rent figures for houses from Tangentyere was unsuccessful. Although Tangentyere knew how much rent it collected in total from town camps, it did not allocate this rent definitively to individual houses. Tangentyere collects rent from individuals under a policy which is in fact more like an informal income tax regime, specifying that individuals with particular levels and types of employment and income will be required to contribute certain levels of payment. Some interviewers did obtain a rent figure for the dwellings they enumerated. How these relate to Tangentyere's rent policy and compare with its administrative figures for total rent collection will be interesting to see. The comparison may be a quite useful check on data quality, showing either that the information recorded accords generally with Tangentyere administrative records or that it is somewhat at odds with those records.

The age question did indeed often initially stop people in their tracks. But when rephrased in terms like 'well, in what year were you born?', it often then elicited an answer. I did however witness one miscalculation by ten years from the year of birth back to the actual age. So perhaps an ability to record the raw answer of the year of birth given may be useful. Or, as Morphy (this volume) suggests, an ability just to put people in five or ten year age groupings.

The income question appeared to cause no great problems, though there did seem to be little if any attention paid to the 'before tax' qualification. Figures quoted were most- 91 - frequently fortnightly social security payment amounts, but were sometimes weekly CDEP pays which could be fairly easily doubled up and put into a broad income category.

Question 37 on the SIPF about whether the person had looked for work in the last four weeks did indeed often elicit the equivocation suggested. Some people were inclined to say 'no' at first, suggesting there wasn't any appropriate work around for them. But then they did wonder whether they were supposed to say 'yes', like on Newstart payment forms. In many ways the distinction between being unemployed and not being in the labour force, which this and subsequent questions try to construct, is of little if any social relevance in discrete Aboriginal communities in sparsely settled northern and central Australia. A person in these communities is either employed or they are not. So the line of questioning is a little bit socially irrelevant and nonsensical.

A couple of other questions which caused confusion also deserve to be commented on. Question 5 on the SIPF, the first that actually needed to be asked, caused confusion. It asked whether the person being interviewed was 'more closely related to anyone else here in this house' than person 1. The relationship between the person being interviewed and person 1 had been established in Q. 4—a question which did not in fact need to be answered, but had been information transferred from the SIHF. Clearly Q. 4 had to be revisited and the answer to it re-identified, if Q. 5 was to make any sense. However, even if this was done, Q. 5 still did not work very well, as there was not necessarily one single other person in the house to whom the person was more closely related. Is a spouse, a child, a parent, or a sibling one's closest relation? Or are they all contenders for that claim?

The question on ancestry (Q. 13) on the SIPF also confused people as it seemed in this context simply to repeat Q. 10 on Aboriginal or Torres Strait Islander origin. Question 10 itself was handled quite well, but like others where the answer was obvious, such as Q. 11 and Q. 12 on whether your mother and father were born in Australia, it was often handled in a joking manner. A joking approach to questions because of their obviousness was common, and was legitimated by the use of this technique in the training video for interviewers.

Question 23 on post-school education, as noted above, often elicited various responses about low level training courses which people had undertaken. This then tied people into answering four more questions about the name of the course, what it involved, the institution at which it was studied and the year it was completed. Whether this was the sort of information that these questions were intended to elicit, again raises the issue of the social relevance of some standard questions to the circumstances of Aboriginal Australians in discrete communities in sparsely settled northern and central Australia.