2050.0.55.002 - Position Paper - ABS Review of Counting the Homeless Methodology, Aug 2011  
ARCHIVED ISSUE Released at 11:30 AM (CANBERRA TIME) 02/08/2011  First Issue
   Page tools: Print Print Page Print all pages in this productPrint All  
Contents >> Contents >> Methodological review consultation / submission process



The ABS initiated the methodological review of Counting the Homeless (CTH) by engaging with a range of stakeholders, including researchers and the homelessness services sector, and with the advice of a Steering Committee comprising representatives from the Australian Government Department of Families, Housing, Community Services and Indigenous Affairs, the Australian Institute of Health and Welfare, and from three states (New South Wales, Queensland and South Australia) represented on the inter-jurisdictional Housing and Homelessness Information Management Group (reporting to the Housing Ministers' Advisory Committee).

An initiating review workshop was held on 21 October 2009, with Professors Chamberlain and MacKenzie, and representatives from Homelessness Australia, as well as from Commonwealth, state/territory and local government organisations. The nature of the ABS’s concerns with the Counting the Homeless 2006 methodology were outlined by the ABS in Issues in estimating the number of homeless in Australia: A paper to inform a review of Counting the Homeless methodology, which was made available on the National Homelessness Information Clearinghouse website in October 2009, and submissions were sought.

Submissions were received from government organisations, academics and eight homelessness services sector organisations. Workshops to progress the review, which involved Professors Chamberlain and MacKenzie, were held in May 2010 and October 2010.

The ABS's initial findings from the methodological review were published on 31 March 2011 in the Discussion Paper: Methodological Review of Counting the Homeless, 2006 (ABS cat. no. 2050.0.55.001). That Discussion Paper announced a public submissions process over the following three months and a series of advertised public forums in each capital city.


The Discussion Paper: Methodological Review of Counting the Homeless, 2006 (ABS cat. no. 2050.0.55.001) was designed to initiate discussion with key stakeholders and interested members of the public on the review's initial findings.

To maximise exposure to the findings, the advertised public forum details were also emailed directly to many stakeholders, and the forums were held in each capital city through April and May 2011. Over 150 people attended the public forums, and many more attended sector or jurisdictional specific discussions with the ABS. The ABS was also invited to participate in discussions about the review findings in a range of meetings.

The structure of the discussion forums was designed to both explain the proposed changes to homelessness estimation methodology as a result of the review of Counting the Homeless (CTH) and seek feedback from stakeholders on both the estimation approaches and wider issues in homelessness measurement. As the Discussion Paper was the first time that the CTH methodology had had its assumptions tested, the discussion forums were expected to help inform stakeholders both of the nature of the CTH estimates and of the differences proposed with the reviewed estimates. It was expected that these forums would support the submission process by bringing both the estimation issues and the wider homelessness measurement issues to the fore.

However, some of the complexities in the CTH methodology were new to most stakeholders, and when the detailed information on the Counting the Homeless methodology was coupled with the new ABS material testing the CTH assumptions, in some of the forums there was insufficient time for attendees to further discuss their thoughts and opinions about the proposed changes. To help capture that input as early as possible in the consultation period, forum participants were encouraged to put their issues in writing to ABS as early as possible, even if their final submissions were not lodged until much later.

Following the public discussion forums, and prior to any submissions being received, ABS met in late May 2011 with 24 representatives from Homelessness Australia in a facilitated discussion about the review findings. At that meeting, Homelessness Australia and ABS agreed that they had a common goal to develop good estimates of homelessness. There was also a common recognition of the importance of having an understanding of services experience in order to understand the complexity that is homelessness, and in turn to understand appropriate methodological approaches and to assess aspects of the quality of any estimates that are produced. Other key points from the meeting included:

  • recognition of the limitations in every existing dataset for use in the estimation of homelessness;
  • that to better understand flows and the trends in repeat periods of homelessness, the duration of spells of homelessness etc, it was important to have more than just point-in-time (prevalence) measures, and the wider ABS work program on homelessness statistics underway or planned was noted;
  • a wide range of measurement issues, which were also picked up in submissions and are discussed below under the theme headings or discussed in separate Key Issues sections of this Position Paper;
  • general agreement that there needs to be some additional research into understanding the size of the homeless youth population, including any under-enumeration in the Census, in response to which ABS is proposing a quality study of homeless school students;
  • the need for Homelessness Australia and the ABS to work together in the future, including through the newly formed Homelessness Statistics Reference Group, to develop a richer and more reliable information base for analysis and reporting on homelessness in Australia; and
  • ABS noting the need to finalise, by late 2011, any decisions on a reviewed estimation methodology to be used with 2011 Census data (and backcast to 2001 and 2006 for trend estimation). ABS and Homelessness Australia agreed to use this time to give due consideration to the comments provided throughout the consultation process, including for areas of measurement where there was no immediate solution to issues of underestimation. The need for more time to reflect on and address the issues raised in consultation was a common theme throughout the consultation process.

Main themes from consultation

There were common themes in the statistical issues raised during the various consultation phases, from the initial workshop in late 2009 through to the submissions in response to the Discussion Paper. The statistical issues are brought together and discussed under the theme headings below and in the separate forthcoming second edition of this Position Paper.

Another issue raised by some homelessness services stakeholders was a concern about whether new estimates of the number of people enumerated in the Census who are likely to have been homeless on Census night, which are lower than the estimates included in the CTH report, may result in a reduced focus on, or funding for, homelessness services. ABS noted that a prevalence (point-in-time) measure, while useful for monitoring trends is only one aspect of the complexity that is homelessness. The ABS Discussion Paper provided a possible range for a relatively large estimate of the number of people experiencing homelessness during a year, which could provide the focus for understanding another dimension of the scale of the issue. ABS household surveys will commence reporting official statistics on this measurement perspective from late 2011. This and other ABS homelessness work program initiatives will report further on homelessness to improve measurement and reporting beyond a five-yearly prevalence estimate at the time of each Census. Together these data developments are expected, over time, to improve the evidence base for decision making.

Actions following consultation meetings

During the consultation meetings the ABS agreed to:
  • undertake further analysis and provide additional data to attendees of the discussion forums to further test the revised methodology. The additional material despatched to consultation participants included estimates for smaller geographic regions (Statistical Local Areas, and Statistical Subdivisions); and Age and Indigenous status disaggregated by State/Territory; No usual address; Need for assistance; and Indigenous status disaggregated by age;
  • establish a Homelessness Statistics Reference Group (HSRG) with representation from the sector, policy agencies and the research community that will guide the wider ABS homelessness measurement work program, including estimation from the Census. The HSRG is now established and will meet in the coming months;
  • publish any changes to estimation methodology that are developed through ongoing consultation and advice from the HSRG, including any implications of harnessing other data sources to re-estimate earlier point-in-time estimates from the 2001 and 2006 Censuses;
  • secure funding for a proposed quality study into homeless school students. The funding has been agreed;
  • publish a position paper one month after the completion of the post Discussion Paper submission period to provide a clearer statement of where the ABS already acknowledges shortcomings in the use of the Census for the purposes of estimating the homeless, as well as making it clearer that there are significant data gaps, and that further work will be required to produce estimates that can be relied upon for reporting the scale of homelessness in Australia, and to measure trends. In progressing that further work, the ABS will work with the sector and others, both bilaterally and through the new HSRG.


The publication on 31 March 2011 of the Discussion Paper: Methodological Review of Counting the Homeless, 2006 (cat. no. 2050.0.55.001) invited submissions on the review findings. Up until 27 July the ABS had received 35 submissions. The submissions were lodged by local councils and state government departments, homelessness peak bodies, homeless services providers, researchers and unions. While the main themes from the submissions have been captured in this paper, the ABS is still reviewing the submissions for any additional implications they may have on future methodology, including the availability and potential of some additional sources of data that were identified in the submissions. ABS will separately follow up with the providers of those submissions.

Overall, the themes in the submissions were consistent with those raised at the discussion forums, but provided more detail. For example, some of the submissions discussed the specifics of the adjustments made in deriving the reviewed estimates published in the Discussion Paper, and provided different opinions on whether a change was required, and if so, what the new methodology should be. The ABS will draw on the expertise of the Homelessness Statistics Reference Group in making its decisions for a final reviewed methodology.

Main themes from submissions

General comments about the methodology

The submissions supported rigorous high quality estimation of homelessness, and highlighted the importance of consistent, repeatable and transparent estimates. One submission noted that a peer review of the CTH methodology was long overdue. Some submissions requested a continuous quality assurance plan be implemented to continue to improve and maintain high quality estimates.

The majority of submissions requested that the ABS more clearly acknowledge, in the documented methodology, where underestimation is likely (for example for youth, people fleeing domestic and family violence, and Aboriginal and Torres Strait Islander people). Submissions also noted the ABS should clearly acknowledge, in its methodological documentation, that the Census does not collect data on, nor derive variables that directly report on, lifestyle, choice or preference; that these choices may impact on the concept and measurement of homelessness; but that these aspects are inferred through a number of other variables collected on the Census (such as home ownership or income).

Many submissions felt that the Review was too focussed on identifying the possible overestimates in the CTH methodology and was not sufficiently focussed on identifying possible underestimates including for the broad areas of youth, people fleeing domestic and family violence, and Aboriginal and Torres Strait Islander people. A couple of submissions felt that the methodology should allow overestimation of some groups to compensate for the underestimation of other groups. This is in contrast to the Discussion Paper's argument that those groups which are overestimated are unlikely to have the same demographics, geographical locations or trend direction as those who have been underestimated, and that aggregate and explicit confidence bounds on broad estimates were more transparent than hidden but presumably offsetting errors in both directions. One submission put forward a view that 'undercounting' can only be addressed in the enumeration stage of the Census and cannot be addressed in the estimation stage.

No submission put the view that the ABS should not produce homelessness estimates from the Census. Many commented that the Census was currently the best source on which to base a point-in-time estimate of homelessness in a comprehensive way.

One submission supported the removal of imputed records from the 'Persons in improvised dwellings, tents, sleeping out' category of homelessness where no one was actually enumerated there on Census night but records were imputed in the Census. No submissions disagreed with the removal of these records.

Many submissions noted that academics and experts in the field should evaluate the methodology underlying the reviewed estimates. As noted elsewhere in this Position Paper, the ABS is continuing to review the estimation methodology and has already identified changes that will need to be made. These changes and any other suggestions will be picked up in ABS work program that is taken to the new Homelessness Statistics Reference Group which has representation from academics and other experts.

One submission proposal, reiterated in several other submissions, was for the review period to be extended until the end of 2012 to allow sufficient time for the HSRG to convene and effectively contribute its advice to ABS before reviewed estimates for 2001 and 2006 and new estimates for 2011 are finalised. ABS has noted that it will receive ongoing advice from the HSRG and will receive, at any time, submissions or advice suggesting improvements to methodology. However, any suggestions that require active interrogation of 2011 Census records during processing to improve their use in final homelessness estimates would need to have been received and considered by the end of 2011 to be certain that any changes could be implemented in estimates for 2011. The 2011 estimates will be published as soon as possible after the second release of Census variables are available for analysis. A longer time frame is possible for proposals that involve only output variable analysis, and the methodology can still be investigated and improved after 2011 estimates are released in the second half of 2012. Such a process of ongoing review and improvement is likely to be sparked by users analysing the 2011 results, by other research, and by analysis of other new ABS homelessness datasets. The ABS will publish, in May 2012, an official estimation methodology, and recompile 2001 and 2006 estimates.

Several submissions also sought more information to explain and justify the differences between the CTH and Review methodologies.

Improved enumeration

Overall, the submissions recognised the importance of accurate enumeration of homeless people in the Census to improve the accuracy of homelessness statistics derived from this source. The ABS does implement strategies to maximise the enumeration of the homeless and other special interest groups in the Census, but looks forward to working further with homeless services and other stakeholders to further improve these efforts in future Censuses.

Submissions provided the following recommendations for improving the enumeration of homeless persons in the Census:
  • the development of homeless-specific Census questions to directly capture homelessness, or any questions that may aid with the correct classification of those who are homeless;
  • work further with the appropriate services to improve the enumeration of homeless people, including the correct reporting of no usual address. Of particular interest was further work to understand and enumerate key population groups such as Aboriginal and Torres Strait Islander people, youth, those fleeing domestic or family violence, and those from non-English speaking populations;
  • contacting local councils and service providers (as proposed by ABS during consultation forums) to provide lists of legal, and where available, illegal boarding or rooming houses to improve the correct identification of boarding houses. One submission raised the challenge of collecting complete and up-to-date boarding or rooming house lists from local councils. The ABS sought such lists from state and territory departments for the 2011 Census (and some have already been provided), as well as from services during consultation on the review, and will consider further expanding the request for lists from local councils and services for the 2016 Census.

Definition of homelessness

On balance, the submission process has surfaced significant interest in a review of the definition of homelessness used in Australia, whether or not that review would confirm the continued use of a cultural definition (with or without adjustments to its current application), or lead to the development of a new definition. ABS will include this review in its work program and progress the issues through the new HSRG.

Many submissions argued that residents in transitional housing management units (or equivalent), which are included in both the CTH and the ABS reviewed estimates, should be retained in estimates of the homeless population, given that these residents do not have security of tenure.

Some submissions supported continuing to use the 'cultural definition of homelessness' as the framework in which to estimate homelessness within Australia. Some of these submissions argued that the cultural definition should be strictly adhered to, and stated a view that the ABS had altered the cultural definition of homelessness, in particular by expanding the recognised exceptions to the definition.

Some submissions called for the definition of homelessness to be reviewed, and that a review be informed by international examples of definitions.

One submission from a service provider argued that people living in boarding houses should not be included in the definition of homeless, as their inclusion in homelessness estimates dilutes the literal meaning of being 'home-less' and that the continued inclusion of boarding houses fails to recognise changes in the standards of residential accommodation.

Some submissions requested that the tertiary homeless category be expanded to include usual residents in caravan parks. This group is excluded from the CTH definition of homelessness but reported separately as a marginally housed group. Others would like to see the scope of the marginally housed group expanded to incorporate new population groups.

Supported accommodation for the homeless

There was general support for the approach adopted by the ABS in the reviewed estimates of using Census data to report on homeless persons staying in supported accommodation for the homeless (previously known as SAAP). To maintain that approach, there was strong support for the existing 'list' and 'green sticker' strategies to be continued. The submissions also urged the ABS to continue to confront the Census data with the data generated from administrative systems on the use of supported accommodation for the homeless, as collected by the Australian Institute of Health and Welfare.

The ABS had noted during consultation that it understood that hotels and caravan parks used by crisis support services, via brokerage, to provide crisis accommodation were not captured in former administrative data on people accommodated in SAAP on Census night, and that the Census offered opportunities to report these people within a supported accommodation category of homelessness. ABS will undertake further research on possible data sources and estimation methods for 2001 and 2006 estimates, and consult on the outputs from the new AIHW Specialist Homeless Services that may support estimation for 2011 and future Census cycles.

New category – 'Persons staying in other temporary lodgings'

There was widespread support for the creation of the new category in the reviewed estimates of 'Persons staying in other temporary lodgings' which reclassified these people from the CTH Boarding house category (tertiary homeless population). These people were reported in the Census without a usual address and were staying in public hospitals, private hospitals, hostels for the disabled etc. There was one request to expand this new category to include persons who were in correctional facilities.

One submission argued that this new group should be included under secondary homelessness rather than tertiary homelessness.

Some submissions argued for people for whom no usual address was reported and who were staying in caravan parks on Census night to also be included in the new 'Persons staying in other temporary lodgings' category. These people are currently classified as staying with other households.

Boarding houses

Several submissions argued that the rules applied to classify dwellings as boarding houses risk missing homeless students, particularly international students.

A few submissions argued that the CTH criterion, which uses the number of unrelated adults (5 or more) as a first step in determining whether a property enumerated in the Census as a private dwelling should be reclassified as a boarding house, should be revised down to 4 or more persons to align with the definition in the Victorian Residential Tenancies Act 1997, which provides for "a building in which there is one or more rooms available for occupancy on payment of rent..." and in which ".. the total number of people who may occupy those rooms is not less than 4..."

The rule of 5 or more, used in part in Counting the Homeless methodology to reclassify as boarding houses those properties that were enumerated as private dwellings, was adopted because the private dwellings with fewer unrelated usual residents were most likely to be share houses (CTH 2008). If the property has been enumerated as a private dwelling, the Census variables cannot directly identify whether any person is renting a room only (which is a critical part of the legal definition of a rooming house in Victoria). If the dwelling was enumerated as a non-private dwelling such as a boarding house, but with less than 5 usual residents, it would be in scope of the boarding house population.

One submission argued that where a property has been enumerated as a private dwelling and the usual residents of that dwelling reported that they rented the dwelling from a parent or relative of one of the usual residents, that property may nevertheless actually be a boarding house and therefore the usual residents may actually be homeless. The review concluded that while this is a possible situation, it is much more likely that the Census collector correctly enumerated the property as a private residence.

A number of submissions advised that boarding or rooming houses are often managed by real estate agents. One service provider submission estimated that in their catchment area, 20% of boarding houses were managed by real estate agents (although it was not clear to ABS if these properties were clearly boarding houses that might be enumerated as such by a Census collector, or would present as a private dwelling and be enumerated as such). The submission therefore argued that it was possible for properties that were enumerated as private dwellings and paying rent to a real estate agent to actually be boarding houses, and the reviewed estimates should consider these properties if the other new rules adopted for the 2006 CTH applied. There were less than 400 people Australia-wide in the CTH boarding house estimates of persons enumerated in private dwellings for the reason that they were renting from real estate agents. Over 85% of these people were in 4 bedroom (50%) or 5 bedroom (35%) properties. The review criterion appears to appropriately distinguish these properties, enumerated as private dwellings with usual residents, as most likely to be share houses rather than a boarding house, although it is possible that some may be boarding houses. Further research and reporting through the HSRG will be undertaken to refine this analysis further.

The rules used to estimate the homeless population in boarding houses are the most complex within the CTH methodology. The reviewed estimates have made the application of some of the rules more consistent, and also looked in more detail to test the assumptions in that methodology. However, as noted in the Discussion Paper, ABS had not been able to test the veracity of all of the dwellings classified by the Census collectors as boarding houses and which were included in the CTH estimates. In this context, several submissions queried the 902 homeless youth aged 12 to 18 years that remain in the reviewed estimates as staying in boarding houses in comparison to other homeless groups. ABS has examined some of the properties that housed these youth on Census night and concluded that some at least are not boarding houses but are lodgings for secondary school students studying away from home e.g. the dwellings contain only secondary school aged youth who are attending secondary school, and adults who are child care workers. The ABS will continue its analysis and take its findings for refining the rules for identifying boarding houses to the HSRG.

As noted above under Definition of homelessness, one submission from a service provider argued that persons in boarding houses should not be included in the definition of homeless. Similarly, as noted under Improved enumeration, some submissions referred to improving lists of boarding houses.

Natural disaster areas

Some submissions highlighted the need to consider how the recent disaster events (floods, fires and cyclones) will be considered in the 2011 estimates. A number of submissions have raised concerns as to whether the rules applied to identifying homeless people in improvised dwellings will exclude those living in temporary housing as a result of recent disaster events.

One submission suggested that this group should be reported on separately to understand both the impact of natural disasters on overall homeless figures, and the changes/ trends in relation to areas not affected by natural disasters.

The ABS is developing a strategy to analyse and report on homelessness in areas affected by recent disaster events which it will take to the HSRG.


The Discussion Paper noted that ABS had not been able to establish any reliable way of estimating homelessness among youth staying with other households and for whom a usual address is reported. This has been the greatest area of concern raised in both the consultations and the submissions. The submissions indicated that the low number of homeless youth staying with other households did not concord with their knowledge about youth homelessness, especially when comparing the numbers 'couch surfing' to the relatively high estimate of youth in boarding houses (although, as noted above under Boarding Houses, that estimate has been inflated by the inclusion of school students boarding away from home in special accommodation for that purpose).

The consultation since the publication of the Discussion Paper has not identified any new data source or estimation method to address this acknowledged underestimation for youth homelessness. The ABS is therefore developing a small and targeted quality study of homeless school students after the 2011 Census to help understand the possible level of homelessness, and in particular, inform on how this is manifested in Census reporting. If successful, the study will also scope a possible methodology for a more frequent nationally representative survey of homeless school students, subject to the ABS obtaining funding to run such a survey.

Many submissions welcomed the quality study into school student homelessness, while others felt that it would not be sufficient to understand the extent of youth who were homeless on Census night but for whom a usual address was reported. One submission supported the quality study proposal but recommended expanding it to other education institutions such as Technical and Further Education, Higher Education, Adult Community Education, as well as to Apprentices and employment services. These institutions are not in scope of the planned 2011 quality study but will be considered as part of longer term work program planning.

A number of submissions requested that Chamberlain and MacKenzie's National Census of School Students (NCHSS) be reinstated, or that a similar survey be conducted to understand the level of underestimation of young homeless people. The ABS understands that the NCHSS was conducted by Professors Chris Chamberlain and David MacKenzie at the same time as the 2001 and 2006 Censuses, as part of their Counting the Homeless methodology. However, it has never been connected to, nor is it part of, the official Census of Population and Housing. The ABS has not had any role in the funding, design, development or conduct of the NCHSS.

The ABS has concluded that the nature of the estimates reported from the NCHSS have a number of significant flaws that preclude their use in addressing the underestimation of youth homelessness from the Census: the NCHSS is based on school reports which capture youth homelessness over a week rather than on Census night; it uses an undercount adjustment methodology that is inconsistent between Censuses; and, it applies an extrapolation to the non-school youth population that is both inconsistent with the stated methodology and which overstates the estimate.

One submission supported the decision of the ABS not to utilise the NCHSS estimates but encouraged the ABS to further consider the 12-18 year age group reviewed estimates, and if there is further evidence available, refine the reporting on the scale of the possible underestimation. This work will be taken to the HSRG.

Several submissions referred to the definition of 'youth', noting that the youth services population can be 12-24 years old or 12-25 years old. While the Discussion Paper adopted the CTH definition of 'youth', it also reported on the 19-24 year age group. The wider 12-24 year age group are reported alongside estimates of 12–18 year olds in the Key Issues section on youth.

Many submissions supported the commitment that the ABS made at the discussion forums to analyse in more depth the 40,000 youth who were visiting on Census night and for whom a usual address is reported. This analysis was to help understand whether there were likely subgroups of visitors where the 'hidden" homeless youth might be most likely to be included. The ABS has commenced this analysis, some of which will be presented in the second edition of this Paper on 5 August 2011. The ABS will continue to undertake further analysis to take to the HSRG for discussion. However, most submissions noted that the most likely opportunity to improve the estimation within the Census data was through improved enumeration, and ABS has been working with homeless services providers in each state and territory to encourage accurate reporting of 'no usual address' by all homeless people, including those who are couch surfing and young people.

Aboriginal and Torres Strait Islanders Peoples

There was a general recognition in the submissions of the challenges associated with the enumeration and estimation of homeless Aboriginal and Torres Strait Islander people. This was highlighted in the Discussion Paper by the low proportion of Aboriginal and Torres Strait Islander peoples who were identified as homeless (i.e. no usual address) and staying with other households on Census night. Many submissions agreed with the Discussion Paper's articulation of Indigenous understanding of Census usual address questions and overcrowding as both potentially masking homelessness.

Recommendations from the submissions included further research to understand Aboriginal and Torres Strait Islander peoples homelessness and its capture within the Census, and the utilisation of experts in Aboriginal and Torres Strait Islander Australian's homelessness to improve the enumeration and estimation of this population. ABS noted during consultation that it was planning focus group work among Indigenous communities on these issues, both to shed more light on the Census analysis but also to support development of culturally appropriate ABS homelessness modules for incorporation in ABS nationally representative Indigenous surveys. ABS research in this area will be taken to the HSRG.

New migrants and culturally and linguistically diverse populations

Many submissions did not agree with the assumptions that new migrants (those who had arrived in Australia some time in the 7 months before the Census and were not from countries from which they were likely to be entering Australia on humanitarian visas) were unlikely to be homeless and hence not included in the homeless estimates. The Discussion Paper noted that the largest group (around 450) of these new migrants were from the United Kingdom (UK) and most were in family groups, with about 30% of the UK new migrants being under 15 years of age. While about a third of the adults from the UK reported nil or negative personal income (potentially non-working spouses), another quarter reported personal incomes of $1,000 or more per week. Overall, about 60% of all the adults were staying with other households reporting their household incomes of $1,000 or more per week. The review had concluded that the reasons of mobility/change of address for reporting no usual address, according to the design and intent of the question, were more likely to apply, on average to these populations, than was homelessness for this group, even though some may have been homeless on Census night.

Submissions also noted that new migrants can have difficulty obtaining secure accommodation, and were particularly concerned about New Zealand migrants who do not require a visa to work in Australia. There were a little over 150 migrants from New Zealand (NZ) in this category under the review, and only about a sixth of the adults reported personal incomes of $1,000 or more per week and over a quarter reported nil or negative income. However, in terms of the household incomes of the people these NZ visitors were staying with, 10% reported negative household incomes and 70% reported incomes of $1,000 or more per week.

Some submissions also reported concern about migrants living in overcrowded and marginal housing.

A couple of submissions requested that the ABS recognise the multi-dimensions of homelessness in terms of the ethnicity of homeless people. While these groups are not discussed in detail in the Discussion Paper, estimates are presented by broad country of birth classification, and the data can also be analysed by Year of arrival, Ancestry, English proficiency, Country of birth of father and mother, and Main language other than English spoken at home. Future reporting of official estimates with a finalised methodology will include the richness that Census data can support.

Marginal residents of caravan parks

As noted above under the Definition of Homelessness, some submissions requested that marginal residents of caravan parks now be included as homeless. There were also requests for more research on the type and length of tenure of those living in caravan parks etc. to understand their circumstances, in order to better understand the strength of any argument to go beyond the CTH conclusion that there wasn't a widely accepted argument for treating them as homeless. Some additional data will be presented in the second edition of this paper.

One submission recommended the use of Tourism Australia data to understand marginal residents of caravan parks.

The ABS has undertaken some additional analysis of residents in caravan parks and will undertake further research to define and measure the marginally housed, and present these findings alongside the homeless estimates to the HSRG.


Some submissions argued that overcrowding is a component of marginal housing that does not meet 'minimum expectations' of housing, and as such should be classified as homelessness. As noted above there were particular concerns that in Aboriginal and Torres Strait Islander households overcrowding may mask homelessness, but it can also be an issue for student households and 'new migrant' households.

The ABS will work to define and report the overcrowded population in more detail, along with other marginally housed populations, and present these alongside estimates of homelessness. The HSRG will consider this issue as part of deliberations on a homelessness definition.

Travellers, including Grey Nomads

Many submissions were concerned that 'grey nomads' were not included in the review homeless population. Some submissions felt that the concept was ill defined and required more research and explanation to assist them to understand this population group. Some additional detail will be reported in the second edition of this paper.

There was broad concern that people who were aged 55 years and over, not employed nor looking for work, travelling with their spouse in a caravan that they own, and staying temporarily in caravan parks may be homeless and not be included in the reviewed homeless estimates as they cannot be distinguished from the 'grey nomads' who were excluded.

Services had reported an increase in older people, particularly women, requesting homelessness services.

Some of the submissions felt that living in a caravan was below culturally acceptable housing and therefore they should be treated as homeless if they reported they had no usual address.

One submission argued that while many 'grey nomads' reported a usual address elsewhere 12 months earlier, and the review had taken this as some evidence to support the anecdotal and tourism reporting of 'grey nomads', this circumstance was inconsequential as to whether they are currently homeless because, for some, homelessness is episodic. Another submission suggested that equivalised income should be used in conjunction with these variables to attempt to identify those who are likely to be homeless. The ABS has identified those classified as 'grey nomads' often had no or low incomes similar to those on pensions, further analysis is presented in the Key Issues section on Travellers. The Census is unable to determine what their source of income is, or the level of their wealth and whether they may be living from their savings. There is anecdotal evidence that travellers seeing Australia are in their initial post-retirement period and travelling for long periods and therefore not having a usual address of six months or more.

The ABS will further analyse, and discuss with the HSRG, rules that may be able to be applied in classifying travellers, including 'grey nomads', that may better differentiate between groups of homeless people and travellers.

Older homeless people

As outlined above in the Travellers section, some submissions were concerned about older homeless people, particularly women, not been adequately estimated in the homeless methodology. They were concerned that 'grey nomads' was masking some of these people who were living in caravan parks.

Improvised dwellings, tents or sleeping out

As outlined in the Discussion Paper, the ABS removed from this category imputed records, as well as those records where the ABS determined they were more likely to be construction workers, owner builders and/or hobby farmers. A couple of the submissions noted that the Census does not collect variables to directly determine the reasons as to why these households were in 'improvised dwellings....'. However the ABS used a number of Census variables to infer this. These variables included occupation, tenure type, employment status, mortgage payments, rent payments and household income. See the Key Issues section in this Paper for more information.

There was little comment in the submissions about removing imputed records. The ABS removed records from the 'Persons in improvised dwellings, tents, sleeping out' where no one was actually enumerated there on Census night but records were imputed for people assumed to live in that dwelling. One submission supported the removal of imputed records in the methodology.

A couple of submissions were concerned that the exclusions did not take account of those who were made homeless due to natural disasters. They may have similar characteristics to those who were identified as owner builders and/or hobby farmers (such as high mortgages on the land). Natural disaster areas will be treated separately by the ABS in analysing results from the 2011 Census (see the section on Natural disasters areas for more information).

There was a view from one submission that those who are living in improvised dwellings such as owner builders, hobby farmers and construction workers were still people who were living temporarily in marginal housing that is sub-standard, and removing them from the homeless population is not counting them as living in sub-standard housing. While the variables cannot determine whether all of this accommodation is sub-standard, the Census collector has determined that they are living in improvised dwellings.

One submission requested that if constructions workers, owner builders and hobby farmers were determined to be homeless after this review, that a separate category should be created in which to track them.

Another submission from a service provider supported the removal from the homeless estimate of people living in improvised dwellings for which there is reasonable evidence they are construction workers or if they had a mortgage on the property.

The ABS will continue to further refine the treatment and reporting of those identified as likely to be construction workers, owner builders and/or hobby farmers in 'improvised dwellings.....', considering the recommendations made through the submissions. Real time analysis of the circumstances of the properties during 2011 Census processing will be one improvement in classification that has not been applied in earlier Censuses.

Domestic and Family Violence and Women's experience of homelessness

A number of submissions discussed women's experiences of homelessness and/or domestic and family violence. One submission argued that the current treatment of homelessness is a male view of homelessness and does not take account of women's homelessness, such as sleeping in their car, or partnering for the night with sex to avoid being on the streets or staying with a violent partner. Some submissions acknowledged the difficulties, raised in the Discussion Paper, in classifying through the Census women and children who are staying with friends or relatives as a result of domestic and family violence but who report a place of usual address, despite not being able to return to that address.

The submissions and the ABS recognise the difficulties both of enumerating and of classifying in the Census output, people who are fleeing domestic and/or family violence. Out of fear, they may not have themselves recorded on a Census form for the dwelling they are staying in, or alternatively they may be recorded as having been at home when they were not actually there, but were staying somewhere else on Census night. For those who are reported on a Census form, they may be reluctant, for a number of reasons including stigma, to report having no usual address on Census night. Alternatively, they may have an expectation that they may be able to return to their home in the future and do not see themselves as not having a usual address.

The advice during consultation was not to assume that women who are fleeing violence are only those with low incomes, and the submissions recommended working with experts to understand further the circumstances of those fleeing violence, and look to ways to improve the enumeration of these people to get a more accurate Census based count of this population. The submissions also urged the ABS to look further into the data to identify whether any more of these women, although reporting a usual address in the Census, can be reasonably classified as likely to be homeless. To date, no rules have been established that could distinguish between women reporting a usual address and who were homeless due to violence, and those women who were visiting other households for social or other reasons such as on holiday on Census night.

Other data sources

There was wide support provided during consultation for additional data sources to be used to understand aspects of the homeless population, and to use them to also better understand the Census data. The data sources mentioned included the ABS General Social Survey (and other ABS surveys that will also include a homelessness module) and the new Centrelink flags for homelessness and at risk of homelessness. The submissions advised that these other sources all have their specific limitations for homelessness measurement, just as the Census does, and these should be clearly acknowledged both in analysis and reporting.

Some submissions made further recommendations for additional data sources that might be used. Examples included using hospital data (to understand both the risk of homelessness, and homeless status prior to admission and on exit from hospitals), Reconnect data, Tourism Australia data and local government or service level data on clients and locations of boarding houses. The ABS will follow up these opportunities as part of its ongoing homelessness measurement work program.

Other areas for further analysis

Some submissions requested that seasonal workers and students should be researched further, due to a concern that there were homeless populations in these groups that may not be adequately captured in current sources and analysis.

One submission highlighted the importance of the interactions between disability and homelessness. While the Census collects data on needs for assistance, this submission felt that this question was insufficient to understand these interactions as it doesn't capture severity or type of disability. The 2010 ABS General Social Survey will provide data on both previous experiences of homelessness and disability status, and future ABS Surveys of Disability, Ageing and Carers will consider incorporating questions on prior experiences of homelessness.

One submission requested a clearer presentation of the differences between the CTH methodology and the reviewed estimate. The Discussion Paper presented both textual discussion of the differences and decision rule trees in an Appendix, but did not clearly show the significance and impact of all of the criteria, such as when 'No usual address' is employed, nor the treatment of some subpopulations (especially children). The submission argued for more consistency with the treatment of 'not-stated' values such as when the dwelling type is not stated, and consistency in treatment of mortgage payments and rent payments, the income thresholds applied and the different employment and student status. The Appendix in this Position Paper presents the information in a different way to improve interpretability of the differences, and ABS will further review rules used in estimation. However some rules are applied for different purposes in different circumstances - for example a higher threshold for mortgage payments is applied when looking at improvised dwellings to better distinguish those who are likely to be owner builders in transition while they build.

Two submissions recommended the use of equivalised household income, instead of household income in the decision rules using income.

Outcomes from the submissions

Following consultation the ABS has determined that it will:
  • utilise the expertise of the homelessness services sector, both bilaterally and through the Homelessness Statistics Reference Group (HSRG), to understand and improve reporting on the characteristics of key homelessness population groups;
  • seek input into an exploration of the definition of homelessness, for initial discussion with the HSRG;
  • continue to articulate and acknowledge possible underestimation, including in the final methodology used for 2011 Census reporting.
  • ensure, through the HSRG, that the explanation of the final methodology for 2011 Census results (and backcasting), and the reasons behind the rules to be applied, are transparent (including for the boarding house classification);
  • continue the current enumeration and estimation approaches for persons staying in supported accommodation for the homeless (previously known as SAAP) on Census night;
  • retain and report on the homeless category for 'Persons in other temporary lodgings';
  • consider the recommendations for improving the enumeration of homeless populations as part of the development of the 2016 Census, and work further with the homelessness services sector in implementation;
  • undertake a quality study of homeless school students after the 2011 Census to provide insight into the likely underestimation of homeless youth using current methodologies. If the quality study methodology proves successful, the ABS would seek funding to run a homeless school student survey on a regular basis;
  • continue analysis of the 40,000 youth who were visitors on Census night and for whom a usual address is reported, and identify whether any subpopulation of this group are more likely than others, on average, to include youth who were possibly homeless on Census night;
  • continue working with homeless service providers in each state and territory to encourage accurate reporting of 'no usual address' by all homeless people, including those who are 'couch surfing';
  • work further on the analysis of the currently classified boarding house population regarding both overestimation and underestimation, for review through the HSRG;
  • for the 2016 Census, investigate local council and other sources for boarding house lists;
  • redefine marginal residents of caravan parks, as part of the marginally housed populations, and present these alongside estimates of homelessness;
  • reanalyse the reviewed estimates for travellers, including 'grey nomads' to assess whether any refinement is required; and
  • review consistency in classification rule for different homeless groups to understand if any changes are appropriate.


The timetable below presents a broad outline of the process of consultation and refinement of the methodology:

31 March 2011Discussion Paper: Methodological Review of Counting the Homeless, 2006 (ABS cat. no. 2050.0.55.001) released
April - June 2011Consultation period
April - May 2011Discussion forums held in each capital city
2 August 2011Position Paper: ABS Review of Counting the Homeless Methodology (ABS cat. no. 2050.0.55.002) released
9 August 2011Census night
August 2011 onwardsCommencement of Homelessness Statistics Reference Group (HSRG)
Working with HSRG and other key stakeholders to develop final methodology
Late 2011Finalise methodology in regard to input processing for the 2011 Census
May 2012Publish official output estimation methodology, and recompile 2001 and 2006 estimates
Late 2012Publish 2011 estimates

Previous PageNext Page