As noted, in the lead-up to the census at Wadeye, the CFO had assessed that an interview team of 20–30 people would be required for the task, assisted by four CCs. This estimate was based on a calculation involving the idea that the CFO and Assistant CFO would be back in Darwin by 8 August for the start of the mainstream census, and that the Wadeye enumeration would be completed fully by then—a proposed enumeration time of eight working days. Against this schedule, given the estimated number of people to be counted and the estimated time required to administer the 55-question IHF, this workforce target was not unreasonable. It was, however, unrealistic under the circumstances.
On the morning of 24 July, the CFO arrived in Wadeye and the TRC Housing Office training room was made available to begin the census process. The first activity scheduled was for the CFO and his Assistant to explain the nature of the task to potential census interviewers, then to provide a day of training on how to administer the census form. Simultaneously, the newly recruited CCs—the seconded TRC employee and a non-Indigenous male who had been resident at Wadeye for a year and employed variously as a casual worker—were to be instructed in their roles. As the CFO’s plan was to be in Palumpa, Peppimenarti and Daly River by 27 July to begin the same procedures there, three days were set aside in order to assemble this workforce, sign them up as ABS casual labour, train them and satisfactorily deploy them to be left in the capable hands of the CCs. This turned out to be an ambitious timetable and one that was dictated more by the pressures on the CFO to administer such a vast census area than by any proper consideration of real training needs.
Almost inevitably, it was nearly midday on day one before a group of interested people finally assembled in the training room. By the time lunch was provided, there were 13 potential CIs present (nine women and four men), but by the time lunch was over only seven people (five women and two men) remained for the training. This manoeuvring continued throughout the next day, leading to substantial turnover during training, involving 22 individuals out of whom only six (four women and two men) finally signed up as census interviewers, although in effect only four participated in enumeration since one of these served as an interpreter for one of the CCs and one withdrew early on. One consequence of this substantial shortfall in labour was that the CCs spent most of their time operating as census interviewers and less time on coordinating and assisting the activities of interviewers. This was unfortunate as the interviewers were deployed singly, and not in pairs, while CC support was physically limited. As a result, any errors or omissions on IHFs had to be discussed and dealt with back at the training room at the end of each day rather than addressed in situ.
Training in the application of the IHF began after lunch on the first day. Taking into account technical glitches and a short break, it lasted for just one and a half hours. It was constructed around a viewing of the census DVD—which was delayed owing to technical difficulties and the fact that many of those present before lunch failed to return—with breaks for discussion. Views on the content and effectiveness of this training were sought from participants afterwards and the following is a precis. The DVD presentation was set in the New South Wales South Coast community of Wreck Bay. One view offered by the local participants was that this setting was too mainstream for the Wadeye context, with interviews conducted indoors in unfamiliar settings (well-furnished lounge and kitchen areas) among small social units. Most participants would also have preferred the presentation and discussion of census content and process to have been conducted in Murrinh-patha rather than English. Although the CFO halted the presentation occasionally to solicit any questions, all this produced was a blanket silence as opposed to the robust discussion in language that one often hears at Wadeye. People had questions, they just didn’t ask them.
As noted, the basic instruction method was to halt the DVD periodically and attempt a discussion of the issues presented. In effect, what this provided was an opportunity for the CFO to establish certain standards in answering particular census questions. For example, it was suggested by one participant that $40 a dwelling should be indicated on the IHF in answer to Question 5 about rent. The Assistant CFO checked this subsequently with the TRC and the idea that each adult paid $10 a week was tabled. The general outcome observed during enumeration was some confusion about the real amounts inserted. This was partly because of the location of the rent question on the census form (at Question 5), which preceded the lengthy business of establishing the real occupants of dwellings including adults (Question 12), from which a retrospective calculation of rent could be derived. Similar conformity was sought for Question 13 on family relationships, though how this was to be achieved remained unresolved. The main advice was that CIs should think in terms of ‘whitefella’ categories, not local ones, and that if individuals were not clear they should defer to the CCs. The problem here was that the CCs had no idea how to translate local family relationships into ‘whitefella’ categories. Instances of such efforts observed during enumeration revealed that individual interviewers worked through the nuances themselves, so any notion that a common understanding prevailed can be discounted.
For Question 12, one of the CCs suggested using TRC administrative data on dates of birth (DOB) in the event that people did not know these—which turned out to be very common. This suggestion was, however, overruled by the CFO as too complex an arrangement, with a preference expressed for acquiring DOB information from the interview. On income (Question 40), the CFO offered to find out standard rates regarding how much people received from various allowances and pensions as well as from the Community Development Employment Projects (CDEP) each fortnight. Although this was never established, the amounts tended to be standardised anyway, with CIs acting intuitively in the field, aided by the grouped nature of income categories on the census form. For Question 46 about the person’s employer, the instruction was to classify any activities associated with the TRC as ‘community services’ even though these could include a diverse range of industry types. Strangely, the DVD content then jumped from Question 25 to 40, so no discussion of the intervening questions was held. Along the way, an ad hoc list of common spellings was compiled for use in the field, although this was rarely utilised in practice.
At the end of the first day, the prevailing view of the training session among local participants was that far too much information was provided in too short a time, with the added difficulty that none of this was in Murrinh-patha. This referred not just to the census content, but to the many administrative details related to interviewer pay and conditions. While in practical terms the effect was high attrition during training, a fair degree of confusion was also evident among those who remained. This all left the CFO and his Assistant somewhat anxious given the time pressure on them to instigate the process and move on to the next community. They were hopeful, however, that a more hands-on approach on day two, with individuals practising by interviewing each other, would help to progress matters. This did, in fact, turn out to be more effective in generating understanding among the six interviewers who persisted through day two. As a precaution, however, the CFO had organised a team of ABS officers to stand by in Darwin to assist with the census if necessary, and arrangements had been made with the TRC to reserve accommodation for them.
One important consequence of ending up with a much smaller workforce than hoped for was a lack of representation from across the local socio-cultural spectrum. Of the four effective interviewers and the translator, all were Murrinh-patha speakers but only one had affiliation with the large Marri Ngarr language group, and none with the other large language group, Marritjevin. Together, affiliation with the latter two language groups encompasses about two-thirds of the population of the Thamarrurr region. While the use of Murrinh-patha as a lingua franca in the region reduced the impact of this shortcoming at the level of communication, the fact that language and clan affiliation often graft on to social groupings and residential location meant that interviewing in certain parts of Wadeye became problematic, if not impossible, for most of the team—especially in light of recent communal animosities. While it is not clear how this affected the population count, it was apparent that interviewers were focused mostly on their ‘own’ areas of town, leaving certain other areas—and therefore particular social groupings—to be covered largely by the non-Indigenous CCs.
Of course, the ultimate consequence of a small workforce was that the count took much longer than was planned for. In fact, rather than taking the eight working days originally hoped for, the enumeration was pursued over 33 working days—from 26 July to 9 September, excluding weekends and two public holidays. Not surprisingly, perhaps, during this extended period the local enumeration team experienced gradual attrition with the final stages of enumeration conducted by just one of the CCs, and ultimately by the CFO and his Assistant. There were a number of reasons for this outcome.
First, the limited training provided for what was an inexperienced team meant that systems were learnt mostly on the job by working through the interview schedule—with mixed results. An important rider here is that none of the interviewers, when questioned, indicated that they had read the Interviewer Household Form Guide or the Working for the Census booklet that were provided with the census bag. The outcome was that while some of the CIs struggled throughout with form completion and administration, others learnt quickly, although all found the task arduous. Part of the issue here was the sheer length of the form with its 55 questions and the need to extract data for every household member. This meant that interviewing was often lengthy, with one dwelling observed taking a whole morning to interview. It should be emphasised, however, that the new, compact, single-form structure of the IHF compared with the two forms previously used did prove highly practical to administer—even in simple ways such as when questioning people under trees or on verandahs in the often blustery conditions of the Top End dry season.
Second, at the end of each day, the process of checking and reconciling the many data items from the front of the IHF with the Interviewer Dwelling Checklist (IDC) and the Master Dwelling Checklist (MDC) was at times chaotic as essential details had not always been completed in the field and forms were sometimes mixed up on return to the Housing Office training room. Third, as they proceeded through the census, most interviewers became enmeshed in social demands from their own family members—such as child minding, card games, shopping, providing meals, visits to the clinic and so on—and the combined effect was a variable rate of progress. The fact that lunch was not provided to interviewers also meant that they tended to stray in the middle of the day and it was difficult for the CCs to gain a regular sense of overall progress.
Finally, conscious of the fact that they were being paid piecemeal rates—$3.29 for each person enumerated—some of the CIs could see that their commitment to the process, over what was turning out to be a longer and more arduous period than expected, was providing diminishing returns to effort, not least because of delays in receiving payment from Darwin, and enthusiasm waned accordingly. On this last point, issues regarding payment for services were something of a running sore from day one. Some of the interviewers had been removed from their standard payroll in order to participate in the census and while ideas were floated early on about the TRC possibly paying them and then invoicing the ABS, the fact is some individuals experienced considerable delay in being recompensed. This led to the unusual situation in which one of the CCs was paying amounts personally out of pocket to be reimbursed later on a visit to the CMU in Darwin.
As mentioned, the population usually resident in the Thamarrurr region is distributed at any one time across a wide area from the east Kimberley and Victoria River valley through to Darwin, though with most found within the region at Wadeye and surrounding outstations. The settlement geography of the Thamarrurr region is shown in Figure 5.1. Most people live in the town of Wadeye but there are some 20 other localities—all outstations—where families also reside, either permanently or occasionally. Most of these have some housing and basic infrastructure, while some have none. As the map indicates, these outstations are located either at coastal sites or on slightly elevated ground above floodplains. Aside from the relative lack of housing and basic services, a major factor that restricts more full-time use of these sites is the poor condition of regional roads and bush tracks, although in August at census time all localities are accessible. Indeed, this is the time of maximum population dispersal across the Top End because of the relative ease of travel. The people of the neighbouring communities of Palumpa and Peppimenarti also have strong social ties with people at Wadeye, and the overwhelming feature of interaction between all of the settlements shown in Figure 5.1 is the constant daily movement of individuals and families between them, as well as beyond the region. For a census that is dwelling-focused, rather than population-focused, this makes counting people a difficult process.
The emphasis in the conduct of the census was to ensure that every dwelling in the region had at least one completed IHF, either with the details of people present or to the effect that the dwelling was unoccupied. This was precisely in line with instructions in the Community Coordinator Manual and the Working for the Census Guide for Interviewers. Accordingly, there was a clear sense in which the count would be considered accomplished at the point when all known dwellings had been processed. Leaving aside some checking of forms by the CFO and CCs to ensure that all questions had been completed, no mechanism was deployed in the field to establish the population coverage of the count in the sense advocated by Martin and Taylor (1996)—again, apparently in line with census instructions, although some confusion reigns here. According to an internal ABS discussion paper on the 2006 IES, before the census forms were to leave communities, CFOs were to check all forms against community lists where possible. This was to help verify counts and coverage, and if insufficient coverage or errors were identified these were to be corrected before returning the forms to the CMU—in this case, in Darwin. This did not occur at Wadeye. Instead, on Saturday 5 August—13 days after arriving in the town—the CFO left for Darwin with all the forms that had been completed up to that point and quality checking was then done at the CMU (see Chapter 7).
One thing that was clear, however, was that the checklist procedure at Wadeye did not work well. As a result, it was difficult in the final stages of the enumeration to know precisely which dwellings had been covered. The basic problem was that CIs often went to dwellings that were different to their allocation off the MDC. By the time the CFO left with all existing completed forms on 5 August, an unknown number of dwellings in town remained to be enumerated (although a figure of up to 20 was mentioned), while several outstations—including Nama, Wudapuli, Table Hill and Fossil Head—had still to be visited. Accordingly, some census forms were left behind with the CCs to continue with the enumeration while assessment of coverage was continued in Darwin by painstakingly matching the existing forms with known dwellings. Instructions for continued census follow-up were subsequently communicated to the CCs, and one of them continued to fill out census forms until the end of August. Final mopping up, however, did not occur until 9 September, when the CFO returned to Wadeye to collect the last of the forms.
On this issue of dwelling coverage, time and again the need for constant local intelligence from outside the census team on the whereabouts of families was underscored. Ideally, this should have been a primary role of the CCs, but their lack of detailed knowledge of the community prevented this. Almost immediately on day one, empty dwellings were encountered because people were away at funerals, had travelled to outstations or the dwelling was due for repair, and it was apparent that tracking occupants in line with the allocation of particular families to dwellings on the MDC was going to be a challenge. In order to gain a sense of the scale of this issue, a quick drive around Wadeye with a key informant on day two revealed that 15 dwellings were empty because their occupants were away from town.
The idea that all dwellings have a single family name associated with them—as per the IDC and the MDC—or the more conceptual notion that individuals necessarily ‘belong’ to particular dwellings was flawed given the prevalence of extended and multi-family or multi-dwelling social groupings involving considerable intra-community mobility. A degree of procedural confusion was also observed here with some interviewers holding the impression that if a person could not be allocated to a lot number they couldn’t be counted. In a number of instances observed, however, it clearly helped that interviewers were sensitive to the prospect of joint family living arrangements and efforts were made to allocate individuals accordingly. Also, what might be thought of as a particular family dwelling did not always turn out to be so. A random comparison of the distribution of families by dwelling between the 2005 Thamarrurr community census results and the 2006 ABS Census indicated that a substantial redistribution within town and away from town had occurred in the intervening months. While some of this occurs anyway for various reasons, the destruction of many dwellings during community violence in 2006 produced considerable upheaval, with a chain reaction of redistribution occurring as families regrouped into safer social clusters and locations.