Table 6.1 presents indirect estimates of early age mortality for the Aboriginal population as a whole. These are based on data on number of children ever born and children surviving collected from all women aged 25–39 years in the 2002 NATSISS survey (Preston & Palloni 1986; United Nations (UN) 1983). As mortality measures obtained through indirect procedures are sensitive to the assumptions on mortality patterns underlying the models—and, in some cases, to methods used in the estimation process—the mortality estimates were generated using three different estimation techniques—the Trussell method, the Preston–Palloni method, and the Palloni–Heligman method—and nine model mortality patterns. These patterns included four from the Coale–Demeny model life table system (north, east, south and west) and five from the UN pattern (Latin American, Chilean, South Asian, Far Eastern and general). For comparison purposes, also presented in the table are probabilities of dying extracted from existing life tables for Indigenous and non-Indigenous Australians.

a. The reference period for the indirectly estimated ₃q₀ values is approximately 1998, while the indirectly estimated ₅q₀ and ₁₀q₀ estimates approximately correspond to the period 1995 and 1992, respectively.

b. From the latest experimental Indigenous life table (ABS 2003a: 81–82).

c. From Kinfu & Taylor (2005). Note that the reference period for all three estimates from the registration data (i.e. ₃q_{0, 5}q₀ and ₁₀q₀) is around 1999.

Source: Author’s calculations from the 2002 NATSISS RADL

The results in Table 6.1 show that the estimated probabilities of dying, generated using alternative assumptions and estimation procedures, fall within a narrow range, which provide confidence on these estimates. These estimates also display a good deal of consistency with the estimates from conventional life tables that are derived from registration data. The probabilities of dying obtained by averaging the results from the different methods and model life table families indicate that the probability of dying before reaching age three was 27.5 children per thousand around 1998, which is fairly comparable with the estimate based on registration data reported by Kinfu and Taylor for the period 1996–2001 (2002, 2005). The estimated under-five mortality rate stood at around 33 per thousand children in 1995, while the probability of dying before age 10 was around 43 per thousand children in 1992. Both of these estimates are higher than the respective estimates for 1999 obtained by Kinfu and Taylor from the registration data, providing some evidence of possible recent decline in child mortality in the population. However, despite this promising trend, there are still significant differences in survival between Indigenous and non-Indigenous children. As can be seen from the table, early age mortality among Indigenous children is three to four times higher than their non-Indigenous counterparts. To reduce—and eventually eliminate—this inequality, it is important to identify the most vulnerable group of the population or isolate the factors to which Aboriginal child mortality is highly responsive. This is the subject of the next section.