There are definitional factors at play in the difficulties encountered by the ABS in enumerating the highly mobile remote Indigenous population. There is the question of how people define ‘resident’ and ‘visitor’; this binary categorisation relies on a settler Australian view of relationship to place that, as the case studies and the observations at the CMU show, is in tension with an Indigenous view that rests on a sense of ‘belonging’ to a particular place. Then there is the question of what to do about ‘residents’ who are absent at the time of the count, and how to decide whether or not they are likely to be counted elsewhere.

In the Alice Springs town camps, there was a big improvement on the situation in 2001. Then, many visitors who were present at census time were nonetheless not counted due to the attempt to apply a ‘usual residents’ or de jure basis for enumeration. In 2006, a de facto approach was adhered to, with the result that all of those present were counted, at least as far as could be ascertained. While the same conceptual approach was adopted at all other sites, some confusion was observed about whether to count certain individuals as ‘present’ and precisely who qualified as a PTA. The observations at the CMU after the count revealed that this confusion was widespread in the Northern Territory. This was partly a consequence of the instruction to interviewers to be flexible and include people who were away but might not be counted elsewhere—a judgment call that was difficult to make at times for interviewers and interviewees. A fundamental difficulty also arose due to the rolling nature of the enumeration over several weeks, as this lent itself to the possibility of individuals—even households—being overlooked altogether or being double (or even triple) counted.

The reasons for and the regional patterning of PTA phenomena in 2006 need to be analysed in detail—centrally and at the CMU level—so that adequate training on how to treat different kinds of cases can be delivered, to the CFOs initially and then on down the line. Training on this issue should have a scenario component, in which people are presented with a range of ‘real-life’ examples to decide on. This should happen at all levels of training. It is hoped that the detailed information contained in our observations will allow the ABS to devise a better articulated strategy for dealing with PTAs in 2011. Our research shows that while mobility is pervasive, it is not random, and it also points to other factors that might reduce the scale of the problem: shortening the time frame of the rolling count and making better use of local knowledge.