Contents >>
Data linking method and quality of data linkage
DATA LINKING METHOD AND QUALITY OF DATA LINKAGE
The 2011 Indigenous Mortality Project used information from deaths that were registered during the Census processing period in 2011–12 when Census name and address were available as linking variables. Probabilistic linking methods were used to bring the datasets together and identify the best match. This process involved comparing name, address and other personal information common to both files, and generated a single numerical measure of how well two particular records matched. Strict quality controls were applied during the linkage process and extensive clerical review was undertaken to maximise the likelihood that links identified in this project were in fact true links that combined the Census and Death registration record for the same individual. Detailed information on linkage methodology is available in Information Paper: Death Registrations to Census Linkage Project — Methodology and Quality Assessment Australia 2011–12 (cat. no. 3302.0.55.004).
At the completion of the linkage, 93% of death registrations had been linked to a Census record. The raw linkage rate for Aboriginal and Torres Strait Islander deaths was 80%, a considerable improvement over the corresponding figure of 74% in the 2006 Census study. While the Census aims to count every person in Australia on Census night, inevitably some people are missed. Results from the Census Post Enumeration Survey were used to calculate an indicative figure of how many death registrations may not have had a corresponding record in the Census file. After applying this adjustment factor, the linkage rate for Aboriginal and Torres Strait Islander deaths would rise to about 90% compared with an adjusted linkage rate of 96% for non-Indigenous deaths. Detailed information on linkage quality is available in Information Paper: Death Registrations to Census Linkage Project — Methodology and Quality Assessment Australia 2011–12 (cat. no. 3302.0.55.004).
CHARACTERISTICS OF LINKED AND UNLINKED RECORDS
The distributions of linked death registrations by age, sex, jurisdiction and remoteness were generally well aligned with those in the total death registration file. There were, however, some differences between linked and unlinked deaths and these resulted in some differences in characteristics between linked and total deaths. For example, Aboriginal and Torres Strait Islander male deaths were slightly under-represented in the linked file compared with the total file. While 52.6% of Aboriginal and Torres Strait Islander deaths on the linked file were for males, 54.6% of Aboriginal and Torres Strait Islander deaths on the total file were for males. Deaths among people aged under 50 years were also under-represented in the linked file. In contrast, deaths for people aged 70 years and over were over-represented on the linked file (33.3% compared with 29.3% in the total file). There was some variation by jurisdiction, but overall the distribution of the linked file was within two percentage points of the corresponding distribution for each jurisdiction in the total file.
Linked and Unlinked Records by relative socioeconomic advantage and disadvantage of area
Linked and unlinked death registrations for Aboriginal and Torres Strait Islander people followed similar patterns of distribution across the deciles of the SEIFA index of relative socioeconomic advantage and disadvantage (IRSAD). For more information on SEIFA see Technical Paper: Socio-Economic Indexes for Areas (SEIFA) 2011 (ABS cat. no. 2033.0.55.001). Although there were some differences between the distribution of Aboriginal and Torres Strait Islander linked and unlinked records in the first, second, third and fifth deciles, overall the linked file was representative of the total file. Non-Indigenous linked and unlinked records were also fairly evenly represented across the deciles.
Linked and Unlinked Records, by Indigenous status and relative socioeconomic advantage and disadvantage of area — 2011–12
|
 | Aboriginal and/or Torres Strait Islander | Non-Indigenous | Total(a) |
|
|
|
|
| Linked | Unlinked | Total | Linked | Unlinked | Total | Linked | Unlinked | Total |
| % | % | % | % | % | % | | % | % |
|
| The Index of Relative Socio-Economic Advantage and Disadvantage (in deciles) | | | | | | | | | |
| First decile (most disadvantaged) | 40.7 | 43.2 | 41.2 | 14.9 | 15.6 | 14.9 | 15.2 | 16.8 | 15.3 |
| Second decile | 13.5 | 14.5 | 13.7 | 14.2 | 13.5 | 14.1 | 14.2 | 13.6 | 14.1 |
| Third decile | 11.7 | 10.0 | 11.3 | 12.2 | 11.6 | 12.1 | 12.1 | 11.6 | 12.1 |
| Fourth decile | 8.3 | 8.0 | 8.3 | 10.4 | 9.0 | 10.3 | 10.4 | 9.0 | 10.3 |
| Fifth decile | 6.1 | 3.5 | 5.6 | 9.2 | 8.7 | 9.1 | 9.1 | 8.5 | 9.1 |
| Sixth decile | 5.7 | 6.1 | 5.8 | 8.7 | 8.2 | 8.6 | 8.6 | 8.1 | 8.6 |
| Seventh decile | 3.9 | 3.9 | 3.9 | 8.2 | 7.4 | 8.1 | 8.1 | 7.2 | 8.0 |
| Eighth, ninth and tenth deciles (most advantaged)(b) | 4.6 | 4.8 | 4.6 | 19.8 | 20.4 | 19.9 | 19.6 | 19.7 | 19.6 |
| Total(no.)(c) | 1 884 | 461 | 2 345 | 140 037 | 10 201 | 150 238 | 142 697 | 10 758 | 153 455 |
|
(a) Includes records where Indigenous status was not stated.
(b) To ensure confidentiality, the eighth, ninth and tenth deciles were combined.
(c) Includes records for which relative socioeconomic advantage and disadvantage of area could not be determined.
More information on the characteristics of linked and unlinked records is available in
Information Paper: Death Registrations to Census Linkage Project — Methodology and Quality Assessment Australia 2011–12 (cat. no. 3302.0.55.004).
Print Page
Print All