Australian Bureau of Statistics
2049.0 - Census of Population and Housing: Estimating homelessness, 2006
Previous ISSUE Released at 11:30 AM (CANBERRA TIME) 11/09/2012
|Page tools: Print Page Print All RSS Search this Product|
FEATURE ARTICLE 2: METHODOLOGY USED TO CALCULATE HOMELESS ESTIMATES
As noted in the above table, imputed records have not been classified as representing homeless people. While no imputation is undertaken for rough sleepers, imputation is undertaken for an improvised dwelling which may have been occupied on Census night and about which the collector was not 'absolutely certain was unoccupied' but for which no contact could be made. In summary, in 2006 there were about 12,000 people enumerated in improvised dwellings as either being at home or with no usual address, from which about two thirds were classified as not being homeless based on the rules for estimating homelessness, and about 1,200 imputed records. If the imputed records reflected a cross section of the people enumerated in improvised dwellings then about two thirds might be expected to represent people who were not homeless. However, the rules applied to estimate the homeless population from the improvised dwelling count require the use of tenure, rent and mortgage payments, and labour force status, to eliminate the mobile construction workers, hobby farmers, owner builders etc.. None of these variables is available for imputed records.
There were 659 records imputed in improvised dwellings as either being at home or with no usual address where there was no corroborating information that the dwelling was occupied on Census night, but also no strong evidence that it was not occupied on Census night. There were further 512 imputed records where no form was collected but a count of the numbers of people in the improvised dwellings was obtained by the collector from a third party. Together these imputed records account for about 10% of those enumerated on forms in these dwellings. This imputation rate is more than double that of the national average rate of imputation, and suggests that the nature of the dwellings for which imputation was undertaken is that they are used for temporary, intermittent or periodic occupation for work or on weekends or holidays and they are not the usual residence of homeless people. The imputed people may not exist, or they may be counted elsewhere at their primary residence.
While it is possible that a few of the imputed records do relate to a person who was homeless on Census night, we know neither their characteristics nor their location. Arbitrarily including a few imputed records adds nothing to the analysis or understanding of this homeless population and potentially skews the resulting distributions of characteristics.
For more information on analysis of this group see the ABS publications: Discussion Paper: Methodological Review of Counting the Homeless, 2006 (cat. no. 2050.0.55.001) and Position Paper - ABS Review of Counting the Homeless Methodology, August 2011 (cat. no. 2050.0.55.002).
PERSONS IN SUPPORTED ACCOMMODATION FOR THE HOMELESS
The ABS considers that the vetting process to allocate people to the short supply of Supported Accommodation Assistance Program (SAAP) accommodation is unlikely to provide supported accommodation for those people who have accommodation alternatives, and that most supported accommodation does not provide the ABS definitional element 'security of tenure'.
To estimate the number of people in supported accommodation for the homeless on Census night in 2006, ABS first included all persons in dwellings identified by the Census Area Supervisors and collectors as non-private dwellings and classified by the building owner / manager as 'hostels for the homeless, night shelter, refuge'. ABS then added people who were in dwellings flagged as being supported accommodation using the Census list and green sticker strategies.
Imputed records are included in the estimates for this group because of the certainty of the nature of the dwellings flagged either by ABS staff or by jurisdictions or services, and the irrelevance of the personal characteristics of occupants to ascertain homeless for people staying in SAAP properties.
Some supported accommodation, such as some transitional housing, may not meet the ABS definition of homelessness because the tenants have security of tenure in the dwelling. However, information in the Census cannot distinguish the transitional housing properties that have security of tenure. The ABS will investigate the possibility of separately identifying such properties with security of tenure in future Censuses.
(To estimate the number of people in supported accommodation for the homeless on Census night in 2001, although the list and green sticker strategies were used, the data was not retained, and data from the Australian Institute of Health and Welfare (AIHW) Supported Accommodation Assistance Program (SAAP) collection on the number of clients and accompanying children accommodated on Census night were used. See Explanatory Notes for more information.)
Rules for estimating Persons in supported accommodation for the homeless
The following table presents the rules that are applied to classify the persons who were most likely to be homeless on Census night and staying in supported accommodation for the homeless. The rules start with the Census non-private dwelling category of 'hostels for the homeless, night shelter, refuge' and adds persons who were in a dwelling flagged as being supported accommodation through the 'list' and 'green sticker' strategies.
PERSONS STAYING TEMPORARILY WITH OTHER HOUSEHOLDS
This group covers people for whom the elements of the ABS definition of homelessness that are lacking are 'security of tenure of the dwelling' and 'control, or access to social relations'. Visitors have no tenure. If they also, on balance, are most likely to lack accommodation alternatives then they are classified as homeless. Groups such as grey nomads, construction workers, recent migrants, other travellers are assumed, on balance, to have accommodation alternatives.
The homeless operational group 'Persons staying temporarily with other households' commences with the analysis of persons enumerated in any private dwelling structure (except an improvised dwelling, tent, sleepers out) and who reported having 'No usual address'.
This group will contain people who were visiting on Census night for a range of reasons including those who are 'couch surfing'. It is unlikely that all of the people enumerated as 'visitors' without a usual address being reported on Census night meet the definition of homelessness.
People in private dwellings who report no usual address fall into two groups:
Given the differences in the living circumstances on Census night of the two groups of people described above, each group needs to be considered using different judgements to determine whether the persons in the group are, on balance, likely to be homeless. Some of the people in 'visitor only households' will be families moving to a new location for work, people who have recently moved to, or returned to Australia and have not, or will not be living in their current property for six months or more in the Census year (given the Census takes place in August). In these circumstances they have correctly answered that they have no usual address, but they are not homeless.
Many other people in this visitor only household group were people travelling on Census night. Of these, the ABS classified one group as 'grey nomads' who were unlikely to be homeless and therefore were not included in the homeless population. Grey nomads are defined as people in dwellings where all people in the dwelling reported no usual address, were aged 55 years and over, were not in the labour force, and were staying in caravans, cabins or houseboats on Census night. The phenomenon of people travelling around Australia in their retirement is growing, and in 2006 the great majority of these 'grey nomads' were enumerated in holiday destinations in the northern beaches in New South Wales, in Queensland, Northern Territory and northern Western Australia.
Another group of people in 'visitor only households' were staying in properties (other than caravans, cabins or houseboats) where the property was owned outright or on which they reported mortgage repayments. They were not staying with friends or relatives. These people were not included in the homeless estimates as the judgement was made that they reported no usual address because they were either travelling or moving primary residence, and staying in their holiday or second home at the time of the Census.
Another group of people in visitor only households were staying in properties (other than caravans, cabins or houseboats) that they were renting. These people were not staying with friends or relatives and their characteristics suggested that they were travelling rather than being homeless on Census night.
The ABS also classified as not homeless people who were, on balance, most likely to be new migrants to Australia or returning to Australia to live. They were people who reported being overseas in August 2005, and who were renting or occupying premises on Census night and reporting no usual address. For new migrants the year of arrival was 2006. On balance, most of the people within this group were unlikely to be homeless and had just not yet had the opportunity to settle in or back into Australia, or if they arrived in July or August would not be able to occupy their current address for six months in the year of the Census.
The ABS recognises that there are a number of groups that would be underrepresented in this homeless group of 'persons staying temporarily with other households'. These include youth, people fleeing domestic and / or family violence and Aboriginal and Torres Strait Islander Australians who may be homeless but for whom a usual address on Census night is reported. They cannot be separated from those people who were visitors on Census night (such as those on family holidays, visiting relatives etc.). This 'invisibility', for some, of the homeless condition in the Census dataset should be taken into consideration when using the data for this group.
ABS has not yet been able to implement any reliable way of estimating homelessness among youth staying with other households and for whom a usual address is reported in the Census. Service providers and researchers indicate that the low estimates of homeless youth staying with other households does not concord with their knowledge about youth homelessness. Guided by its Homelessness Statistics Reference Group, the ABS is continuing to undertake research and development to improve the estimation of homelessness, including youth homelessness. In particular, the ABS has been undertaking a quality study to inform the potential development of a nationally representative homeless school students survey.
In 2006 there were 1,928 imputed records for 'visitors' with no usual address, which are not classified as homeless. In reviewing the reported visitors with no usual address in visitor only households, 86% were determined, on balance, not to be homeless based on the detailed characteristics of the groups that can be studied among these households that suggests more plausible interpretations of the reporting of no usual address. Many of the dwellings housing 'visitor only' households for whom a usual address was not reported were in holiday areas (e.g. the snow and coastal Queensland). These geographical areas may have many dwellings for holiday makers as well as second residences. These dwellings could look occupied on Census night but may not have been because the owners were in their primary residence on Census night, or because the holiday dwelling was not rented out on Census night but was at other times during the Census period. Any records imputed based on the selection of a responding property in a similar location is likely to be at least as likely (86%) to generate a false signal for homelessness as determined for the reported visitor only households. While there are some imputations of households with both usual residents and visitors with no usual address, again the locations of these properties and the higher than average likelihood of non-contact in these areas suggest the visitor imputations are unlikely to reflect homelessness. On balance there is no strong evidence that the imputations reflect homelessness.
More analysis on these groups can be found in the previous ABS publications: Discussion Paper: Methodological Review of Counting the Homeless, 2006 (cat. no. 2050.0.55.001) and Position Paper - ABS Review of Counting the Homeless Methodology, August 2011 (cat. no. 2050.0.55.002).
Rules for estimating Persons staying temporarily with other households
The following table presents the rules that are applied to classify the persons who were most likely to be homeless on Census night and staying temporarily with other households. The rule starts with the broad Census dwelling category of persons enumerated in a private dwelling (which was not classified as an improvised dwelling, tents, sleepers out) and who reported no usual address. The category is then refined for those groups of people who were unlikely to be homeless on Census night among two broad groupings:
PERSONS IN BOARDING HOUSES
The element of the ABS definition of homelessness that people in boarding houses are assumed to lack is 'control of, and access to social relations', and their staying in boarding houses is assumed to reflect a lack of accommodation alternatives which together indicate homelessness. Note that halls of residence for students, dwellings for members of religious orders and institutions such as hospitals are not boarding houses for homeless people.
The homeless operational group 'Persons in boarding houses' commences with analysis of persons enumerated in non-private dwellings that were classified by their owners / managers as 'boarding houses, private hotel'. It is expected that in most cases these boarding houses etc. have been correctly classified as by the owner / manager. However, because the Census boarding house category is not designed to approximate homelessness, it will include some student halls of residence that are not part of any one school but serve multiple schools in a region. Rules are used to exclude from homelessness estimates those dwellings that, on balance, appear to be more likely to be a student hall of residence.
The rules also aim to reclassify other non-private dwellings as boarding houses where the characteristics of the people in the dwelling suggest that they may be housing large numbers of homeless people and can be assumed, on balance, to more likely to be a boarding house although they were classified by the owner / manager as 'staff quarters', 'hotel, motel, bed and breakfast' or as a dwelling type that was not classifiable, or a classification was not stated.
The rules also attempt to reclassify any dwellings that were identified by Area Supervisors and Census collectors as private dwellings but the characteristics of the occupants are such that, on balance, they are more likely to be a boarding house rather than a large (apparently) unrelated group household. If a dwelling is a private dwelling, but operates illegally on the basis of multiple room by room tenancies, the Area Supervisor and collector may enumerate it according to its legal presentation rather than on the basis of its unobserved operation as a boarding house. The rules are applied to identify these dwelling amongst large (apparently) unrelated group households. However, this method is limited and is likely to overestimate the number of large (apparently) group households as boarding houses due to the limitations of the Census in capturing all relationships in the household. The Census only captures the relationships to the first person on the household form (and child relationships to person 2), so if people in the household are in couple relationship but neither of the couple is reported as person 1, then such households with five or more usual residents may be incorrectly classified as group households rather than the family household it really is. That same household may then also be incorrectly assumed to be a boarding house and its residents classified as homeless. The correct classification of households as family households can also be affected when there are persons who were temporarily absent (PTAs) on Census night.
Rules are applied to these large (apparently) unrelated group households to ensure that, on balance, households such as student households, retirement villages, nursing homes, homes for the disabled, convents / monasteries and other religious institutions are not incorrectly reclassified as boarding houses for the homeless.
The same rules for 2001 and 2006 are applied to classify boarding households as homelessness. However, for 2006, additional information available from this Census about need for assistance with core activities and volunteering is used to refine the rules to assist with ensuring the best possible reclassification of private dwellings as boarding houses. The aim of using these variables is to correctly classify dwellings that were more likely to be convents or monasteries (and other similar religious institutions) as well as for dwellings more likely to be facilities for the aged. The necessary items for this classification refinement were not collected in the 2001 Census. The rules for 2001 therefore omit these steps which refine the classification of boarding houses.
To determine, on balance, that the dwelling is most likely to have been a boarding house for homeless people, variables such as labour force status, student status, income, tenure type, need for assistance with core activities, religion and volunteering are used. These are shown in more detail in the table below. All of the variables such as income are applied to the individual in private dwellings rather than considering the household income, because if they are in boarding houses individuals would not be sharing their economic resources.
Rules for estimating Persons in boarding houses
The following table presents the rules that are applied to classify persons who were most likely to be homeless on Census night and staying in boarding houses. The rules start with all persons enumerated in dwellings that were classified as 'boarding house, private hotel'. Then additional rules are applied to remove dwellings that are student halls of residences. Rules are applied to 'staff quarters' to pick up any that, on balance, were more likely to be boarding houses. Rules are also applied to 'hotel, motel, bed and breakfast' and 'other and not classifiable' or 'not stated' dwellings to identify if, on balance, they were more likely to be boarding houses. Finally rules are applied to private dwellings which appear to be large unrelated group households to determine if, on balance, they are more likely to be boarding houses.
PERSONS STAYING IN OTHER TEMPORARY LODGING
The homeless operational group 'Persons staying in other temporary lodging includes people who reported having 'no usual address' on Census night who were enumerated in non-private dwellings which were classified by the Census Area Supervisors and collectors as non-private dwellings and also classified by the building owner / manager as ‘hotel, motel, bed and breakfast’. People staying in other temporary lodgings who have no usual address lack the ABS homelessness definition element of 'security of tenure of the dwelling'. The rules then consider the income and employment status of these people to determine whether they could be considered to be homeless based on a lack of accommodation alternatives.
This group will differ from the estimates provided for this group in the Review estimates that the ABS published in Discussion Paper: Methodological Review of Counting the Homeless, 2006 (cat. no. 2050.0.55.001) because a number of non-private dwelling categories were included in the preliminary estimates that are specific exclusions from the ABS statistical definition of homelessness. As outlined in the ABS definition, they are excluded because although their living conditions lack one or more of the key elements of 'home':
The specific exclusions are for:
For the same reasons, people in non-private dwellings which were classified as 'other welfare institutions' were not considered to be homeless and not included in this homeless operational group. Other welfare institutions include institutions providing mainly residential care for adults in need of care such as benevolent homes and drug / alcohol rehabilitation centres.
Rules for estimating Persons staying in other temporary lodging
The following table presents the rules that are applied to classify the persons who were most likely to be homeless on Census night and staying in other temporary lodgings. The rules start with the persons enumerated in the non-private dwelling category of 'hotel, motel, bed and breakfast' who reported no usual address and is refined to remove people who were, on balance, unlikely to be homeless on Census night.
PERSONS LIVING IN SEVERELY CROWDED DWELLINGS
People in severely crowded dwellings are considered to lack the 'control of, and access to social relations' element of the ABS definition of homelessness. Lack of accommodation alternatives is assumed for people living as usual residents in a severely crowded dwelling. The judgement is that people who had accommodation alternatives would not remain in such severely crowded circumstances.
The homeless operational group 'Persons living in severely crowded dwellings' is operationalised as those people usual residents of a private dwelling that, according to the Canadian National Occupancy Standard (CNOS), requires four or more extra bedrooms to accommodate the usual residents.
The concept of crowding is based upon a comparison of the number of bedrooms in a dwelling with a series of household demographics such as the number of usual residents, their relationship to one another, their age and their sex.
There is no single standard or measure for housing utilisation, however the CNOS is widely used internationally and the ABS uses it for its measures of crowding for other purposes. It is a suitable standard for use with Census data because all of the required variables for its calculation are available from the Census, although family coding limits the suitability of it in large complex family households, and where persons may be temporarily absent on Census night.
The CNOS is sensitive to both household size and composition. The measure assesses the bedroom requirements of a household by specifying that:
Households living in dwellings where this standard cannot be met are considered to be overcrowded.
There may be some underestimation associated with the application of the CNOS to Census data. It is not possible to create a CNOS estimate of the number of extra bedrooms needed for households where any key piece of information is missing. This includes the number of persons per dwelling, age of the persons, the relationship in household, in some cases, where at least one person (who is not the spouse of person 1) is temporarily absent on Census night. In addition, there may be cases where usual residents are not recorded on the Census form due to fear by the residents that they may be found to have more residents living in the dwelling than are allowed by their lease agreement.
Treating overcrowding as homelessness at this severe level of lacking four or more extra bedrooms is designed to prevent the misclassification of people as homeless who may choose to live together under some crowding to save money, to be close to family, or for other reasons etc.. In addition, it also takes account of the limitation of the Census household form which only seeks relationship information within the household in relation to 'person 1', as well as child relationships to 'person 2'. This limitation of Census family coding results in misclassification of family relationships, particularly for large households with complex family relationships or households which contain multiple families, or where persons are temporarily absent. Households that look like crowded group households in the Census may actually include a number of couples. Under the CNOS a single adult requires their own bedroom but a couple can share a bedroom, and the masking of relationships can inflate the crowding measure.
In the ABS Survey of Income and Housing (SIH), which does not go to very remote Australia, there were no dwellings sampled which required four or more extra bedrooms in either 2005-06 or 2009-10, and only a few in each of 2003-04 and 2007-08. The Census data presented over 11,000 non-Indigenous people in private dwellings requiring four or more extra bedrooms in 2006. And for people in dwellings requiring three extra bedrooms, the Census estimate for non-Indigenous people in private dwellings is 20% above the SIH average for the four survey cycles quoted. Drawing the line at four or more extra bedrooms required in the Census data aligns with the concept of extreme or severe overcrowding, and reasonably well with results from ABS Indigenous surveys, and avoids overestimation from Census data at lower thresholds due to family coding, persons temporarily absent and potentially errors in the numbers of bedrooms being reported.
Other Census variables are not applied to imply accommodation alternatives in severely crowded dwellings. While some of the residents may own the dwelling in which they live in severely crowded conditions, the presence of other people that contribute to crowding demonstrates loss of the power to exercise control and choice in this living situation.
Rules for estimating Persons living in severely crowded dwellings
The following table presents the rules that are applied to classify the persons who were, on balance, most likely to be homeless on Census night and live in severely crowded dwellings. The rule starts with the persons enumerated in private dwellings where the dwelling would require four or more extra bedrooms under the CNOS and then refines the group by removing people who have already been considered as homeless in the other homeless groups.
OTHER MARGINAL HOUSING GROUPS
The ABS will present, alongside the homeless estimates from the Census, estimates of the number of people in selected marginal housing circumstances. These are limited to categories that can be obtained from the Census. Other marginal housing, such as housing with major structural problems or where residents are in constant threat of violence, cannot be obtained from the Census and are therefore not included.
These marginal housing groups not only provide an indication of the numbers of people living in marginal housing close to the boundary of homelessness, but can also provide a possible indication of people who may be at risk of homelessness.
Three marginal housing groups are classified from the Census:
Each of these groups are discussed below.
PERSONS LIVING IN OTHER CROWDED DWELLINGS
Persons living in other crowded dwellings are those usual residents living in dwellings reported in the Census where the dwelling requires three extra bedrooms to accommodate them according to the CNOS (see Glossary and the above section for more details on 'Persons living in severely crowded dwellings'). Under the operationalisation of the ABS definition they are not classified as homeless but are considered to be in marginal housing and may be at risk of homelessness.
Rules for estimating Persons living in other crowded dwellings
The following table outlines the rules used to estimate the number of persons living in other crowded dwellings and ensures that no person who has already been counted as homeless is also counted in this marginal housing category.
PERSONS IN OTHER IMPROVISED DWELLINGS
Persons in other improvised dwellings are those people who were enumerated on Census night in the dwelling category of an improvised dwelling, tent or sleepers out who reported either being 'at home' on Census night or having no usual address, and are not considered, on balance, to be homeless (see above on 'Persons in improvised dwellings, tents, sleepers out'). Under the ABS definition they are not classified as homeless because they have accommodation alternatives, in many living situations, the dwelling will be adequate. They are, however, included in marginal housing and may be at risk of homelessness.
Rules for estimating Persons in other improvised dwellings
The following table outlines the rules used to estimate the number of persons living in other improvised dwellings and ensures that no person who has already been counted as homeless or in other crowded dwellings is also counted in this marginal housing category.
PERSONS MARGINALLY HOUSED IN CARAVAN PARKS
Persons marginally housed in caravan parks are those considered to be in marginal housing and at risk of homelessness. However, not all persons living in caravan parks are considered to be marginally housed. For example, those living in cabins will have access to their own kitchen facilities and bathroom. However, the Census data do not allow caravans and cabins to be separately identified. Others living in caravan parks on a long-term basis have an element of security of tenure, and for some people they have chosen to reside in a caravan park due to convenience, cost or location and could select other accommodation alternatives.
Persons marginally housed in caravan parks are operationalised as those people who were enumerated on Census night:
Rules for estimating Persons marginally housed in caravan parks
The following table outlines the rules used to estimate the number of persons marginally housed in caravan parks and ensures that no person has already been counted as homeless or in other crowded dwellings is also counted in this marginal housing category.
These documents will be presented in a new window.
This page last updated 9 November 2012