APPENDIX 3 QUALITY OF INDIGENOUS STATUS DATA IN THE 2006 CENSUS
The ABS aims to produce high quality data from the Census. To achieve this, extensive effort is put into Census form design, collection procedures, and processing procedures.
There are four principal sources of error in Census data: respondent error; processing error; partial response; and undercount. Quality management of the Census program aims to reduce error as much as possible, and to provide a measure of the remaining error to data users, to allow them to use the data in an informed way.
When completing their Census form, some people do not answer all the questions which apply to them. In other cases, some people completing a form on behalf of someone else, or completing details on a form on behalf of someone else, will either miss questions or not be able or willing to supply an answer on that person's behalf. In these instances, a 'not stated' code is allocated during processing, with the exception of non-response to age, sex, marital status and place of usual residence. These variables are needed for population estimates, so they are imputed using other information on the Census form, as well as aggregate information from the previous Census.
The processing of information from Census forms is now mostly automated, using scanning, Intelligent Character Recognition and other automatic processes. Quality assurance procedures are used during Census processing to ensure processing errors are kept at an acceptable level. Sample checking is undertaken during coding operations, and corrections are made where necessary.
The Census form may be completed by one household member on behalf of others. Incorrect answers can be introduced to the Census form if the respondent does not understand the question or does not know the correct information about other household members. Many of these errors remain in the final data.
This Appendix is based on the 2006 Census Data Quality Statement for Indigenous Status. More detailed information on data quality is available in the Explanatory Notes in this publication and the Census Dictionary, 2006 (cat. no. 2901.0).
ENUMERATION OF ABORIGINAL AND TORRES STRAIT ISLANDER PEOPLES
As in previous Censuses, the ABS put in place an Indigenous Enumeration Strategy (IES) as part of the collection of the 2006 Census to achieve the most accurate count of Aboriginal and Torres Strait Islander peoples in both nominated discrete communities and elsewhere. This strategy was developed to have sufficient flexibility to allow for the unique cultural aspects of Australian Indigenous societies that could affect the enumeration, and to raise the quality of the count of this small but significant population group. In many nominated discrete communities, collection of Census information was undertaken by an interviewer, using a tailored Interviewer Household Form, with 17% of Indigenous people being enumerated using this method. Non-response to the question on Indigenous status was 0.7% when this approach was used compared with 1.7% overall. In other areas, Indigenous people were enumerated using standard procedures and forms. Special collectors skilled in Indigenous languages and cultures were available to assist in these areas if required.
The question on the standard household form asking about a person's Indigenous origin was moved forward to Question 7 in 2006, from Question 17 in 2001. This was reflected in the reduced item non-response rate for the question (based on all Census forms returned to ABS, i.e. excluding imputed records), from 2.0% in 2001 to 1.7% in 2006, as early questions are more likely to be completed than those appearing later in the form. The item non-response rate in 2006 was higher among people aged 75 years and over (4.7%) and lower (1.4%) for people under 65 years of age. People aged 65 years and over accounted for 29% of total item non-response to the Indigenous status question, although they represented only 14% of the people counted in the Census when a form was returned.
Most of the 5.7% of total Census records with Indigenous status unknown for 2006 is attributable to imputation for persons for whom a completed Census form was not returned. In 2006, 4.1% of total Census records were imputed compared with 2.1% in 2001, more than offsetting the reduction in item non-response for Indigenous status. As noted above, while age, sex, marital status and place of usual residence are imputed for these people, Indigenous status is not. Therefore, these imputed records are not identified as either Indigenous or non-Indigenous in Census counts by Indigenous status. The ABS adjusts for this imputation when producing official population estimates of Aboriginal and Torres Strait Islander peoples. See also Appendix 2: Indigenous Estimated Resident Population - Method of Calculation in this publication.
AREAS AFFECTED BY COLLECTION ISSUES
In the 2006 Census, in some states and the Northern Territory, the number of Indigenous people counted in the Census in some urban areas and regional towns was below what might have been expected based on 2001 Census results and evidence of growth in these areas since the 2001 Census. Increases in the number of non-responding dwellings (and therefore of Indigenous status being not stated in the imputed records for these dwellings) is believed to have impacted on the 2006 counts of Indigenous people in some parts of Australia. For example, there is evidence to suggest that there has been population movement between some of the surrounding communities and Katherine and Alice Springs in the Northern Territory. This is supported by the 2006 Census counts. However, dwelling non-response in these two urban centres was the major contributor to unknown Indigenous status in these areas. In Katherine, where 11.4% of Census records had unknown Indigenous status, 9.9% of records were imputed, and in Alice Springs, where 7.9% of Census records had unknown Indigenous status, 6.7% of records were imputed.
Affected areas appear mostly to be where there were issues with the recruitment and retention of Census collectors (e.g. northern Western Australia and the Northern Territory). These areas tend to have higher numbers of non-responding dwellings, and evidence suggests that these dwellings may have contained higher than average proportions of persons of Indigenous origin. This includes evidence from both state/territory governments and from academic research about the movement of Aboriginal people between Indigenous communities and urban areas, both as a result of changing policies with programs such as Community Development Employment Projects (CDEP), as well as the strong economy in some localised areas providing increased employment opportunities (as alternatives to Census-based employment).
ABS preliminary analysis of 2006 Census data has shown that counts may also be affected in areas where residents living in Indigenous households may be away or difficult to contact during the enumeration period, because of cultural, social or sporting events. Analysis of those persons who were away from Indigenous communities on Census Night suggests that many of these people were not included in the Census count for the Indigenous population. The findings also raise the possibility that the combination of an extended enumeration period in remote areas with a Census Night enumeration in non-remote areas may also have resulted in some persons who were away from home during this period being missed from the Indigenous count.
The ABS plans to do further evaluation work on the quality of Indigenous counts, in particular in urban areas, and on response levels for Indigenous status. Below is a list of SLAs for which Census counts of Indigenous people have declined significantly between 2001 and 2006.
In WA, the following SLAs experienced a decline between the 2001 and 2006 Censuses in their Indigenous population of at least 100 people and representing at least 5% of their population. It is believed that these lower than expected counts of Indigenous persons may be the result of collection issues (including high dwelling non-response):
- Broome (S), located in the Broome IREG - Indigenous population declined by 622 people. The Indigenous status question was not stated for 14.2% of people usually resident in this SLA and 12.6% of the count was imputed into dwellings for which no form was received.
- Halls Creek (S), located in the Kununurra IREG - Indigenous population declined by 426 people. Indigenous status was not stated for 6.3% of people usually resident in this SLA and 5.6% of the count was imputed into dwellings for which no form was received.
- Port Hedland (T), located in the South Hedland IREG - Indigenous population declined by 203 people. Indigenous status was not stated for 25.9% of people usually resident in this SLA and 24.8% of the count was imputed into dwellings for which no form was received.
- Ngaanyatjarraku (S), located in the Kalgoorlie IREG - Indigenous population declined by 133 people. Indigenous status was not stated for 1.0% of people usually resident in this SLA and 0.5% of the count was imputed into dwellings for which no form was received.
2006 Census counts for the SLA of Coober Pedy were lower than expected for both the Indigenous and non-Indigenous populations. The Indigenous population declined by 22.3% (78 people) between 2001 and 2006. A higher than expected number of unoccupied dwellings may partly explain the lower than expected count, as it suggests that the usual residents were absent during Census enumeration. Indigenous status was not stated for 16.0% of people usually resident in this SLA, and 13.3% of the count was imputed into dwellings for which no form was received.