Population counts represent the key output from each national census as they establish the base from which population estimates are subsequently derived and the levels at which population characteristics are established. Such information is vital in assessing change over time in social indicators and plays a key role in the assessment of policy impacts. A notable feature of successive Indigenous census counts has been their volatility and unpredictability with numbers invariably greater each time than change due to natural increase alone would suggest. Indigenous population change recorded over the most recent inter-censal period (2001-2006) provides no exception.

While the picture at the national level has been well researched, it is at the local level where programs are delivered and the need for accurate data is greatest. However, there has been little attempt to assess the robustness of population change data at this level. This presentation examines the spatial pattern of change in census counts and assesses these against a range of likely predictors. We find that for some areas the Census provides reasonable estimates, whereas for others the Census is more akin to a sample survey with implications for analysis, interpretation and policy utility.