See Aboriginal and/or Torres Strait Islander Origin (ABLP), Birthplace, Ethnicity.
Random Perturbation of Table Cells
A new method for the random perturbation of census data has been used in 1996. There is a user requirement that there should be one and only one figure for each census characteristic and that the components of each table add to the total of that table but also that the perturbed figures are close to the 'true' figures.
Various methods for confidentialising the data have been investigated and there is no feasible way to satisfy all user requirements. However, as the main problems for 1991 Census data were caused by the aggregation of randomised Community Profile data at the lowest level to form larger geographic units, the ABS has decided to separately aggregate and perturb each geographic level. This means that the perturbed figures for each geographic level are close to the 'true' figures but there is a loss of additivity between the geographic components and their aggregation. For example, data for Collection Districts which are components of a particular Statistical Local Area do not add exactly to the data for the Statistical Local Area. However, it should be noted that the total for the Statistical Local Area is statistically speaking closer to the true population total than the sum of the totals of the component Collection Districts.
To assist clients of the 1991 Census Community Profile Series, a balancing item was provided at State level. This cell contained the true table total so that clients could see the difference between the aggregated total and the actual total.
For the 1996 Census, because of the different method of application of random perturbation outlined above, balancing items are not included with any standard product. Given that random perturbation is applied independently at each geographic level, the ABS does not expect that users of the randomised data are compromised without the true table totals. For those specialist clients who do require them, the ABS provides a customer service whereby customised tables containing true table populations can be supplied, for example, counts of Sole Parent Families by Collection District. Note that such tables would themselves be subject to limited random perturbation. However, in most cases this would have no effect, although in a few cases some very small cells would be affected.
See also Confidentiality, Introduced Random Error.
Recoding is the process of aggregating categories of a classification into groups; the groups may contain the full basic classification, or only part of it. Each of the census data variables disseminated has a basic classification. Output may be produced using this basic classification level, or using recodes.
Tables can be customised to individual requirements using recodes. For example, if data are required for school children only, a suitable recode for the variable Age (AGEP) may show the age groups 5-8; 9-12; 13-15; and 16-18, reducing the table to a more manageable size.
Recoding may also be necessary to meet the confidentiality requirements of the ABS, to prevent the identification of individual persons or organisations. For example, a table showing the full classification of occupation cross-classified with the full classification of Birthplace (BPLP) at Statistical Local Area (SLA) level, would contain many cells with counts of 1 or 2. In such cases some reduction in classificatory detail, or of geographic detail, would be necessary so that the table would be less affected by introduced random error or other measures used to de-confidentialise the data. For example, birthplace may be recoded into 12 groups and occupation into 50 groups.
Due to technological changes, no derived variables need to be pre-defined on the Unit Record File. The software used for the 1996 Census tabulation has the capability to derive new variables efficiently and readily as required during tabulation. For this reason, many derived variables found on the 1991 unit record file are not included in the 1996 unit record file. The advantage to clients of census data is the increased flexibility of deriving variables for use in customised tabulations, using Client Services.
Many recodes have already been designed and incorporated as standards, based on demand from previous censuses. Clients are advised to select the most suitable of these for their purposes if possible. The design and implementation of non-standard recodes for individual requirements is done by ABS Client Services. The ABS encourages clients to use standard ABS recodes to allow meaningful comparison of data.
See also Client Services, Classifications, Confidentiality, Section 1 - 1996 Census Classifications, Introduced R andom Er ror, Mnemonics, Tables, Variables.
Reduced Output Spatial Database (ROSD)
This is the store of digital geographic information used in most census output products, such as CDATA96.
The ROSD was constructed by reprocessing data which were drawn directly from the Output Spatial Database (OSD). This reprocessing reduced the size of the database to approximately 650 Mb. Some additional name attribution was also added to the base map dataset. This name attribution was obtained from, and added by Mapinfo Australia.
The main reason for producing the ROSD was to supply an integrated set of boundaries and base map data for use in the ABS product CDATA96. The ROSD is also an appropriate size for other desk top mapping packages. The major constraint of this task was that the data were to be wholly contained on a single CD ROM. When the ROSD was being developed, the current technology in use for CD ROMs meant that the database had to be below 650 Mb in size.
The ROSD is also the source of the 1996 Census Reduced Digital Boundaries and 1996 ABS/PSMA Reduced Base Map. These are available for sale to the public and provided in CDATA96.
See also Output Spatial Database, CDATA96, Reduction.
Reduction (associated with digital geographic information)
The process of removing some of the detailed data from digital spatial or associated data, so making the data more useful in desktop mapping systems. To achieve this the Output Spatial Database (OSD) was altered in the following three ways:
- several non-essential classes of base map features were removed;
- the number and content of Oracle tables containing attribute and systems data were rationalised - a process generally referred to as thinning; and
See also Output Spatial Database, Reduced Output Spatial Database.
- the number of points representing each linear and area feature, in both the boundary and base map data, were reduced - a process generally referred to as filtering.
The 1996 Census standard reference maps incorporate two sets of maps: the 1996 Census Statistical Local Area (SLA) Maps and the 1996 Census Collection District (CD) Maps. These are paper maps which are printed for the ABS, as required, by the New South Wales Land Information Centre (LIC) at Bathurst.
Note that these are reference maps, which do not contain any statistical data.
Census SLA maps: There is at least one map sheet for each 1996 Census SLA. Each map shows the boundary of an SLA and the CD boundaries contained within the SLA. The name and code of the SLA are included, as well as the CD codes. Some boundary detail for the surrounding SLAs and CDs is also included, as well as selected topographic details.
There are approximately 2,600 SLA map sheets. Each map sheet is printed on an A3 sheet. Some SLAs are too large to fit on a single A3 sheet. The area of these SLAs has been split over two or more A3 map sheets. These map sheets can be fitted together to create a single representation of the SLA. For some maps additional map sheets, containing insets, are supplied. These insets show enlargements of more densely populated areas.
The scale of each map is set so that the area covered by the SLA is maximised on the available sheet area. Maps are therefore not all the same scale. Maps split over two or more sheets are at the same scale.
Census CD Maps: These are the same maps which were used by census collection staff during the collection phase of the 1996 Census. Each map shows an individual CD's boundary and the code assigned to it. Some boundary detail for the surrounding CDs is also included in the map. All census CDs have at least one map sheet covering their area. The maps also show a range of topographic and cadastral features. These features give the boundaries their context and show how they are used to define the CD boundaries.
There are approximately 35,000 CD map sheets. Each map sheet is printed on either an A4 or an A3 sheet. Some CD Maps were originally printed on sheets larger than A3. However, for the purposes of dissemination these sheets have been split into separate A3 map sheets which fit together to produce a single map of the CD.
The CD and SLA maps can be ordered from ABS Client Services and are provided either in person from the ABS or posted direct.
See also Land Information Centre.
See Family, Household, Family/Household Reference Person Indicator (RPIP).
Census Geographic Areas.
Registered Marital Status (MSTP)
This variable records an individual's current status in regard to a registered marriage, i.e. whether he/she is widowed, divorced, separated, married or never married. The partners in a registered marriage must be of the opposite sex as same-sex relationships cannot be registered as marriages in Australia. Registered Marital Status (MSTP) is coded from Question 6.
A question on Registered Marital Status has been asked in all Australian censuses. In 1986, 1991 and 1996 this question provided responses for the categories:
Marital status is applicable to people aged 15 years and over. Note that the category 'married' was called 'now married' in censuses prior to 1986.
See also Marital Status, Married - Registered, Social Marital Status (MDCP).
- Separated but not divorced;
See Other Related Individuals.
See Other Related Individuals.
Question 5 on the 1996 Census Household Form asks for each person's relationship to Person 1/Person 2. The categories provided allow for husband or wife, de facto partner, child, stepchild, brother or sister, unrelated flatmate or co-tenant, or other relationship as specified. Children can be children of Person 1 or Person 2 only , or of both Person 1 and Person 2.
See also Family, Family/Household Reference Person Indicator (RPIP), Household, Household Head, Relationship in Household (RLHP).
Relationship Between Families (FRLF)
This family level variable classifies the relationship between the primary family and the second or third family enumerated in the same household. To identify the second and third families, the variable Family number (FNOF) is required.
See also Family.
Relationship in Household (RLHP)
This is a key variable at the person level. It is used to determine familial and non-familial relationships between persons residing within the same household. Persons who are usual residents of another household are included in this classification but information collected from them is not used in family coding. Theyare assigned to a separate category of visitor.
A question on relationship has been included in all Australian censuses. Prior to 1981, the question asked about relationship to household head. Public reaction against the concept of household head caused the question to be altered in 1981 to relationship to Person 1. Since the 1986 Census, relationship to Person 1 and/or Person 2 has been asked in respect of each child. This was necessary to classify stepchildren.
For information regarding the categories contained within RLHP see the following entries:
Same-Sex Couple, Lone Parent, Married - Registered, Child under 15, Other Related Individual, Non-Dependent Child, Dependent Student, Non-Family Member, Lone Person Household, Group Household.
See also Family Type (FMTF), Family\Household Reference Person Indicator (RPIP).
Relationship in Non-Private Dwelling (RLNP)
This variable identifies whether people enumerated in Non-Private Dwellings (such as motels, hospitals, colleges etc.) are staying there as either:
- members of staff of the accommodation (e.g. owner, proprietor, porter, cook, teacher, warden or family of owner of staff); or
As people in Non-Private Dwellings are enumerated using Personal Forms instead of Household Forms, no information on family relationships is available.
See also Dwelling, Type of Non-Private Dwelling (NPDD).
- residents, guests, patients, inmates etc.
See Other Related Individual.
A question on religious denomination has been included in all Australian censuses, but answering this question has always been optional. The option not to answer this question is provided for in census legislation.
Responses to the religion question are coded to the standard classification of religious groups in Australia.
Data on religion are used for such purposes as planning educational facilities, aged persons' care and other social services provided by religion-based organisations; the location of church buildings; the assigning of chaplains to hospitals, prisons, armed services and universities; the allocation of time on public radio and other media; and sociological research.
RENT (weekly) (RNTD) and (RNTD01)
This variable records the individual dollar amounts of rent paid by households on a weekly basis for the dwelling in which they were enumerated on census night, including caravans etc. in caravan parks. The categories range from $0-$9,999 in single dollar amounts.
The Census is the only source of rent data for small areas and for small groups of the population. Such data are important for housing policy-making and planning, and for studying the housing conditions of minority populations.
A derived variable RNTD01 is also available which groups rent into categories,
e.g. $0-$49, $50-$99, $100-$149 etc.
See also Household, Housing Loan Repayments (Monthly) (HLRD), Tenure Type (TEND).
See Rent (Weekly) (RNTD), Tenure Type (TEND).
See Dwelling, Household, Usual Residence.
Residual Categories and Supplementary Codes
Residual categories in a classification are labelled Not elsewhere classified (nec), Not elsewhere included (nei), Other or Miscellaneous.
These categories are necessary because, although in a classification, meaningful categories are created through the application of certain criteria, not all observations can be classified into a homogeneous group, or the size of the observations does not allow them to be separately identified. For example, in the classification of languages, the minor group 'Chinese' is composed of six distinct languages and one residual category:
The residual category is needed because the six distinct languages do not encompass all the known Chinese languages. The remainder of observations which can be classified as 'Chinese languages' are grouped together in 'Chinese languages, nec'.
Supplementary codes (often called dump codes) are used to process inadequately described responses. Not Further Defined codes (sometimes referred to as Undefined codes) are used to process incomplete, non-specific or imprecise responses which cannot be coded to the most detailed level of a classification, but which nevertheless, contain enough information to allow them to be coded to a higher level of the classification structure. For example, birthplace responses relating to places which cannot be identified as lying within the boundaries of a country separately identified in the Australian Standard Classification of Countries for Social Statistics (ASCCSS), but which lie wholly within the boundaries of one of the classification's Minor Groups, are coded to that Minor Group.
Not Further Defined codes consist of the code of the higher level category to which they relate, followed by an appropriate number of zeroes. For instance, in the above example, responses are allocated a Not Further Defined code consisting of the two digit code of the ASCCSS Minor Group followed by '00'.
It is important to note the distinction between Not Elsewhere Classified categories and Not Further Defined codes. NEC categories are a formal part of a classification's structure, designed to make a classification complete and exhaustive of all observations in scope. Adequately described, specific responses are coded to NEC categories in instances where a suitable substantive category is not included in the classification. As explained above, NFD codes are designed to facilitate processing by allowing inadequately described or non-specific responses to be coded to a broader level of the classification rather than be lost altogether. NFD codes are not part of the classification.
Other supplementary codes are also provided in classifications, for operational purposes, to facilitate the coding of responses to:
Chinese languages, nec (includes Hsiang, Kan)
- Inadequately Described - where a response contains insufficient information to be coded to any level of the classification;
- Not Stated - where no response is provided; and
See also Classifications, Input Processing.
- Not Applicable - where the question does not apply to the person and so no response is required (for example, Year of Arrival is not applicable for people born in Australia.
See Accommodation for the Retired or Aged (Cared).
Rooms in occupied private dwelling
See Number of Bedrooms in Private Dwelling (BEDD)
See Reduced Output Spatial Database.
See Dwelling Structure (STRD).
See Section of State.
See Section of State, Urban Centre/Locality.