Time series analysis of these estimated populations is rendered problematic by changes in census geography at the small area level. However, it is possible to adjust Indigenous and non-indigenous ERPs for the whole of the East Kimberley for the period 1981 to 1991 to account for those who did not report their Indigenous status, and then to apply an approximation of census undercount based on the levels suggested by the 1996 and 2001 ERPs. The results of this manipulation are shown in Table 2.3. The same data are shown graphically in Figure 2.1.

With the exception of 1991, this produces a curve of steady upward growth for both populations over the past 20 years. Of particular note is the fact that in 1996 the Indigenous population exceeded the non-Indigenous population (Figure 1). In 1981, the Indigenous share of the resident regional population was 49% and by 2001 it was 54%.

As a result of boundary changes over this period, what is less clear is how the balance of the estimated Indigenous and non-Indigenous populations within the Northern East Kimberley has shifted over time. An area approximating this region can be identified for 1981 using CDs, although usual residence counts were not available at this geographic level in 1981. Nonetheless, Indigenous and total place of enumeration counts are available at CD-level, and these can be used to begin to establish some sense of the population pattern. According to these figures, the Indigenous population counted within the study region (2,294) was only 36% of the total at that time. Of course, many of those counted were temporary visitors to the region, and the vast majority of these were non-Indigenous. If these are discounted, then the Indigenous share rises to 64.8%.

While it is difficult, in retrospect, to assess the quality of census counts in 1981, especially for the Indigenous population, at the very least an attempt can be made to adjust these counts to create 1981 population estimates benchmarked against the regional ERP, in much the same way as has been done for 2001 in Table 2.3. Such estimates for the Indigenous population within each component part of the study region^[2] are shown in Table 2.4, along with calculations of the numeric and percentage change in population numbers to 2001.

A number of features emerge from these data. First, the overall annual growth rate of the Indigenous population in the Northern East Kimberley (3.3%) is slightly above what might have been expected due to natural increase. This suggests that some movement of Indigenous population into the area has occurred over the 20 year period. Second, there is considerable variation in growth between different localities and areas within the region. On the whole, urban centres display the highest rates of growth, at levels pointing to net in-migration. Some communities appear to have remained more or less static in population size, suggesting net out-migration. In terms of redistribution, there appears to have been population loss in the southern part of the region, and population increase in the north-western area. However, this apparent loss in the south may simply reflect the relocation of people into places such as Warmun and Halls Creek.

Mention has been made of the fact that a relatively large proportion of the population present in the East Kimberley at any one time is comprised of individuals whose usual place of residence is elsewhere. At the 2001 Census, just over 3,000 people, or 22% of the enumerated population in the East Kimberley, fell into this category. The vast majority of these, according to the census, were non-Indigenous people. Typically, in a region such as this, most such non-residents would be tourists and others would be temporary workers (Bell and Ward 2000). Some sense of the scale of such movement through the region is provided by data on estimated overnight visitor numbers in East Kimberley towns in 1994–95, with 70,000 reported in Kununurra, 47,000 in Wyndham, and 32,000 in Halls Creek (Kimberley Development Commission 1997: 36).

This addition to the resident population caused by the temporary movement of people into and out of the region is significant in remote areas such as the Kimberley as it more than offsets any population loss due to out-migration (Bell 2001: 13–14). It also adds to pressure on selected local services, whilst at the same time generating extra demand for goods and services and thereby enhancing regional economies of scale. To this extent, temporary migrants form an important element of the regional economy.

By their very nature, though, temporary movers are elusive in the context of formal statistical collection, and this is particularly the case among Indigenous populations in remote regions as they are frequently mobile over the short term (Taylor 1998). While this may lead to undercounting of the population at census time (Martin and Taylor 1996), it can also produce wide discrepancies between client-based population lists (for example, in the form of clinic registers) and official head counts.

For this reason, the Ord Valley Aboriginal Medical Service based in Kununurra maintains records of all its regular Aboriginal clients and it divides them into those who have a home address of Kununurra, and those whose address is via Kununurra, which essentially refers to surrounding (and distant) communities and outstations. In January 2003, the total numbers counted in this way amounted to 1,644 within Kununurra and 197 outside. It is interesting to compare at least the first of these figures with the derived Indigenous ERP for Kununurra of 1,009 as shown in Table 2.4. While direct comparison of these population levels is not possible owing to the different methods upon which they are based, the fact that the health service-based figure is almost two-thirds higher than the census-based estimate suggests that either the census-based estimate is too low, or that the health service-based figure includes individuals that the census might have counted as usual residents of places other than Kununurra. Resolution of such differences is no easy matter, and is beyond the scope of the present study. Suffice to say that such a variation underscores the fact that the numbers actually demanding and utilising services in the region may often exceed official estimates.