Living arrangements of secondary school students - quality study

In September 2012 the Australian Bureau of Statistics (ABS) published the first official statistical definition of homelessness (see Information Paper - A Statistical Definition of Homelessness (cat. no. 4922.0)), followed by an associated new method for estimating homelessness using data from the ABS Census of Population and Housing (see Information Paper - Methodology for Estimating Homelessness from the Census of Population and Housing (cat. no. 2049.0.55.001).

The first official homelessness estimates based on this methodology and derived from the 2001, 2006 and 2011 Censuses were subsequently released in late 2012 (see Census of Population and Housing: Estimating homelessness, 2011 (cat. no. 2049.0)).

It is generally agreed that estimates of youth homelessness derived from the Census are likely to have been underestimated. Nevertheless, ABS homelessness estimates indicate that youth (defined in this paper as persons aged 12-24 years) are over-represented in the homeless population. In 2011, homeless youth accounted for 25% of all homeless persons in Australia, while youth represented only 17% of the total population at that time (see Census of Population and Housing: Estimating homelessness, 2011 (cat. no. 2049.0)).

Homeless youth were most likely to be enumerated in the homeless operational group 'Persons living in 'severely' crowded dwellings' (52% of homeless youth in 2011). This was followed by 'Persons in supported accommodation for the homeless' (21% of homeless youth in 2011). The operational group 'Persons staying temporarily with other households' accounted for only 12% of homeless youth, however these estimates are likely to be particularly impacted by underestimation issues.

Identifying homeless youth staying temporarily with other households (often described as 'couch surfers') rests firstly on them being recorded as present in a dwelling on Census night and, secondly, being reported as having 'no usual address'. However, it appears that a usual address is often reported for such youth, making them indistinguishable from genuine 'visitors'. A usual address may be reported for 'couch surfers' for various reasons, including that the young person does not want to disclose to the people they are staying with that they are unable to go home, or the person who completes the Census form on behalf of the young visitor assumes that the youth will eventually return to their home. Estimates of youth homelessness for the other homeless operational groups are not subject to these specific underestimation issues.

Neither the ABS nor other researchers have yet been able to establish a robust method, using existing data sources, to more reliably estimate homeless youth staying with other households. This is a major concern for policy makers and service providers, given the strong association between youth homelessness and poorer outcomes in adulthood, particularly in relation to education and employment (see Australian Social Trends: Life after Homelessness, March Quarter 2012 (cat. no. 4102.0)).

A number of recent data developments have sought to build the evidence base for developing policy to address and monitor the levels of homelessness, including for youth. These include but are not limited to:

• The establishment of the Specialist Homelessness Services (SHS) collection, compiled by the Australian Institute of Health and Welfare (AIHW) from data on the clients of specialist homelessness agencies. The collection includes information on 'support periods' (including previous periods of homelessness) by demographic characteristics such as age, education and employment status.
   
• The commissioning of the project Journeys Home: A Longitudinal Study of the Factors Affecting Housing Stability, funded by the Department of Social Services (DSS) and run by the Melbourne Institute. The project tracks a national sample of individuals exposed to high levels of housing insecurity to assist in understanding the various factors associated with homelessness and housing stability.
   
• The Longitudinal Survey of Reconnect Clients, run by DSS to evaluate the role of the Reconnect program in building community capacity for early intervention into youth homelessness.
   

Despite these developments, concerns remain over whether youth homelessness, particularly 'couch surfing', is being adequately captured in the various existing data sources, including Census homelessness estimates.

In response, guided by the ABS Homelessness Statistics Reference Group (HSRG), the ABS has undertaken a quality study with two main aims:

• To investigate the feasibility of conducting a national school survey of the living arrangements of secondary school students.
   
• To determine whether such a survey could address limitations in the estimation of youth homelessness, particularly for estimates of 'couch surfing'.
   

The quality study and its outcomes are the subject of this information paper.

This work was jointly funded by the ABS, the Department of Education, Employment and Workplace Relations (DEEWR) and the Department of Families, Housing, Community Services and Indigenous Affairs (FaHCSIA).

Ideally, a survey of homeless youth would include both those attending school and those who have left school. However, the absence of an appropriate survey sampling frame, as is available for schools, makes extending a survey to all youth very difficult. 

This quality study was therefore restricted to focus on developing a methodology that could provide information regarding the prevalence of homelessness within the secondary school student population.

Results from such a survey could be analysed in combination with Census-based estimates of homeless youth, as well as SHS statistics, to provide a more comprehensive understanding of overall youth homelessness.

ABS obtained approval from the New South Wales Department of Education to contact selected schools that were chosen to participate in the ABS Quality Study.

Importantly, informed consent is required from both the participating student and their primary caregiver prior to the commencement of research involving students. In practice, such consent processes rely on either 'active' or 'passive' consent models. Under both models, students and parents are fully informed of the research process. An 'active' consent model requires a record of agreement to participate, while a 'passive' model assumes consent unless the primary caregiver explicitly denies permission for the student to participate in the study. Both active and passive consent approaches were tested in this quality study.

Under an active consent approach, 20 high schools in NSW (in Sydney and in regional areas) were invited to participate in the quality study. Of these, five initially volunteered to participate and the ABS followed up with nominated teachers or school counsellors to:

• compile a list of students whom they believed were either homeless or at risk of being homeless

• obtain consent from a parent or guardian for the ABS to interview the student

• arrange for the student interviews to be conducted at the school with an appropriate teacher or school counsellor available.
   

Among the questions tested in this initial approach were questions about the student's whereabouts on a designated night (which in a nationally representative survey could be used to establish a point in time prevalence measure of homeless school students). As the quality study interviews were being conducted close to the time of the 2011 Census of Population and Housing, the designated night chosen was Census night. The study would therefore not only serve to test the methodology for a prevalence measure, but also provide a window into how information about the usual residence of homeless youth may have been reported in that Census. 

Unfortunately, very few primary caregivers took the steps required to provide explicit consent for the student in their care to be interviewed, with schools confirming that such processes can be burdensome for busy families. Timing of the study in August was also problematic for those students (and their teachers) undertaking trial exams at this time of year. Several of the schools therefore recommended an alternative methodology be considered, including:

• a passive approach for parental consent for student participation in the study

• surveying all students in a selected class (perhaps via an electronic form) rather than selecting only those students whom the teachers considered to be relevant for the quality study.
   

Surveying all students was also expected to reduce the burden on the school administration (and teachers/counsellors) in relation to managing both classes and separate interviews for a subset of students. A wider selection process would also address the risk that teachers/counsellors may not be aware of all the students experiencing homelessness or being at risk of homelessness, and as a result enhance the quality of the information collected in the study. Finally, this process also avoided the risk of potential stigma associated with a targeted selection process.

Following revision, the ABS quality study proposal using a passive caregiver consent approach was approved by the New South Wales Department of Education. The outcomes of this approach are further detailed in the remainder of this information paper.

A letter of invitation was sent to the principals of 42 schools from across New South Wales, with 14 agreeing to participate in this study. While strict statistical sampling methods were not employed, school delegates were asked to select classes that were known to include students with a range of different living arrangements, literacy levels and English proficiencies. 

Parents/guardians of students in the selected classes were informed of the study via a letter and given the option of removing their child from the study. Students were also advised that participation was voluntary and that they could choose not to complete the study on the day of the survey.

All documents distributed to schools, students and parents avoided the use of the term "homeless", given differential interpretation of this concept. This ensured that participation was not impacted by assumptions that the survey was only relevant to students who were "sleeping rough".

The number of classes surveyed per school ranged between two and twelve. A number of schools suggested that the ABS survey the entire school. While this could be an option for a future nationally representative survey, it was not possible within the limited time frame available for the quality study. 

Two self-complete paper questionnaires were used for the passive caregiver consent approach, one to be completed by students and the other by the teachers of the selected classes. Both forms were designed by the ABS with assistance from the ABS Homelessness Statistics Reference Group (HSRG). 

The student questionnaire collected basic demographic information, along with information about where the student had slept on the night prior to the day of the survey. If the student spent the night at a place other than their legal guardian's dwelling they were prompted to select a reason for staying there from a list of options. All respondents were asked to state the number of nights in the last month that they had spent at their “last night's" accommodation. Respondents who indicated they had not spent all nights at the same place in the last month were asked to identify other places that they had slept overnight in the month prior to the survey and the reasons for staying there. The purpose of this approach was to identify students who only occasionally sleep at the home of their legal guardian and who may have experienced a form of homelessness in the previous month.

The aims of the teacher form were twofold. Firstly to obtain information on class demographics, including for students absent on the day of enumeration, and secondly, to obtain teacher views on which students were experiencing a range of certain living arrangements that may be considered, or indicative of, homelessness.

Enumeration was undertaken in July 2012, during the week before and the week after the NSW school holidays break. A key aspect of the study (under both the active and passive consent approaches) was the anonymity of the students surveyed. The names of students participating in the study were neither requested nor collected.

The process of both introducing the study to students and their completion of the paper form questionnaire took about 15 minutes in most cases (slightly longer in classes with high numbers of students who had English as an additional language). A maximum of three classes were surveyed by one ABS officer during a scheduled lesson. However, class disruption was minimised and (the observed) student focus improved when the quality study was completed at the beginning of a lesson period.

At a small number of schools, the delegates planned for the survey to take place in a school hall, with class groups arriving and departing approximately every 30 minutes. This method worked very well in terms of student behaviour and reduced the likelihood of students discussing the survey content.

The findings of the study are intended to assess the feasibility of the tested methodological approach. As such the results presented should not be considered in any way a representative sample of school students and cannot be generalised or used as an estimate of homelessness amongst Australian school students.

For this study students were defined as having had an experience of homelessness if they reported staying at one of the following places for any reason:

• hostel / boarding house

• youth refuge

• street / squatting / car / train etc.

Students were also defined as having had an experience of homelessness if they stayed at any of the following places:

• other relative's place

• friend's place

• shared house

• hotel/motel

• caravan park

• my own place

for any of the following reasons:

• my parent, guardian or other adult relative wasn't able to take care of me

• I don't feel safe in my family home

• I was asked/told to leave

• I don't like parent/guardian's rules / had an argument

• my family has money problems

• I don't want to live with other people staying in the house
   

A total of 1,432 survey forms were completed by in-scope students on the enumeration days. In general, students were able to complete the questionnaire with relative ease. However the category "My own place" was sometimes wrongly interpreted by students as the home they live in with their parent/guardian, rather than a dwelling for which the student had personal responsibility.

In the earlier stages of enumeration, some questionnaire sequence guides were overlooked by some students. This did not result in any loss of information but could lead to respondent frustration at having to answer seemingly irrelevant questions. ABS officers changed their introduction when administering the survey to reinforce the explanation of sequence guides.

Figure 1 shows that of the 1,432 students who participated in the quality study, 22 appeared to have experienced homelessness at least once during the previous month. Of this group, six students appeared to be homeless on the designated night (three were in a refuge and three had 'couch surfed' or 'slept rough'). Among the remaining 16 students (those who had reported an experience of homelessness in the previous month but not on the designated night) the majority had 'couch surfed'.

As noted previously, these findings are presented to allow assessment of the feasibility of the methodological approach and can in no way be used to derive an estimate of homelessness amongst Australian school students.

The teacher form captured basic information about class demographics (year level, age and sex, including for students absent on the day of enumeration). Of the 85 classes from which students were enumerated in this study, 73 teachers (86%) returned their teacher form. There were reported to be 1,634 students enrolled in classes where a teacher form was completed (Figure 2). Teachers reported that a high number of students were absent, with 322 students (20%) absent on the day of enumeration. Nearly one-third of the schools were enumerated during the final week of term and this was identified by teachers as a contributing factor in the higher than usual rates of absences. 

Teachers also frequently commented that they did not know the exact age of all students in their classes. Most either asked the students in the class how old they were (which worked well for those students present, but not for absent students) or chose to fill in the form after class when they could look at the school records to find the ages of students in their class.

There were 25 classes in which at least one student was identified by their teacher as having experienced difficult living arrangements in the last month. No students were identified as having experienced such difficulties in 46 of the remaining classes, while there were a further two classes where this information was not provided.

A total of 95 students overall were identified by teachers as having experienced difficult living arrangements (Figure 2). As noted, a significant issue for the quality study is the relatively high rate of absenteeism at the time of the study. Of the 95 students whom teachers considered to have experienced some difficult living arrangements, 40 were absent from class on the enumeration day, and for a further 25 students the teachers did not confirm whether or not the student had completed a student form.

As noted above, student names were not collected or retained for analysis. Comparison of student and teacher reports is therefore limited to cases where a confident match could be made based on demographic characteristics. 

Examining these cases shows that of the 22 students who were identified on the student form as having experienced homelessness, either on the designated night or in the previous month, only three appeared to have also been identified by teachers as having experienced difficult living arrangements. The majority of students who reported experiences of homelessness came from classes in which teachers did not identify any students as having issues with their living arrangements.

This study has highlighted a number of issues which would need to be addressed in considering a nationally representative study of schools. 

Firstly, this study confirmed the challenges associated with gaining active parental consent for a large scale school collection, particularly in the context of a constrained survey period. While a passive consent process was more successful for this study, and was supported by the schools involved, further consultation would be required to determine the appropriateness of this approach across Australian jurisdictions and school sectors.

As mentioned previously the scope used in this quality study was limited to youth enrolled in a secondary school. The absence of a structured frame outside the school system makes it difficult to extend the scope to all youth. The approach employed in this study fails to provide information on youth aged 15-18 years who no longer attend school, as well as those aged 19-24 years who are past high school age.

Importantly, results from the 2011 Census suggest that homeless youth aged 15-18 years are less likely to be attending school than those who are not homeless. In 2011, 35% of all homeless 15-18 year olds reported they were not attending school compared to 15% of 15-18 year olds who were not homeless. This issue is particularly evident in the operational group 'Persons staying temporarily with other households'. More than half (51%) of 15-18 year olds in this group reported they were not attending school.

The under-representation of youth attending school in this group suggests that a school based approach would likely continue to underestimate youth homelessness, particularly for 'couch surfers'.

The results of the study also showed that high rates of absenteeism among potentially homeless students impacts on data quality. Overall, a high number of students were absent across the schools, with 322 students (20%) reported by teachers as absent on the day of survey enumeration. The timing of the study may have contributed to this result as nearly one-third of the schools were enumerated during the final week of term and this was identified by teachers as a contributing factor in higher than usual rates of absences. 

However, it is also conceivable that students experiencing difficult living arrangements may be more likely to be absent from school. This is reflected in the teacher results, with 42% of the students that teachers identified as experiencing difficult living arrangements being also reported as absent on the day of the student collection. 

To maximise student participation and school capacity to support the survey, the timing of a future survey would need to not only avoid the busy times in schools around assessment tests, but also avoid the beginning and end of terms to limit the number of absences due to holiday arrangements of families. 

An appropriate method for collecting information to determine whether absent students were likely to be homeless would also be important to consider in any future survey. 

Any survey of this type is potentially affected by non-sampling error associated with issues such as recall limitation, differential interpretation of survey questions, sensitivity to questions and the impact of social desirability. Given potential stigma associated with experiences of homelessness, results from the current study may be impacted by reduced willingness of students to report genuine experiences of homelessness. In addition, some teachers involved in the study also reported lacking some of the knowledge required to complete an accurate assessment of the living arrangements of students. 

The lack of alignment between teacher and student reports is an important result. In this study, the majority of students who reported experiences of homelessness came from classes in which teachers did not identify any students having issues with their living arrangements. This is particularly important to consider given the high rate of absenteeism in this survey, as these students would not necessarily be identified on teacher forms.

Development of a national survey would require further testing with both students and teachers to further test terminology and refine questionnaires to ensure the most accurate reporting through survey forms.

This study confirmed the feasibility of collecting information about student living arrangements using a survey of secondary school students and teachers. However this paper has outlined a range of issues in regards to the methodology used in conducting the study that would require significant improvements before a national survey could be implemented. 

Given the extremely low response rate using an active consent model, the consent model used for a national survey would need to be considered. In addition, the significant issue of student absenteeism would need to be addressed and an appropriate follow-up process for absent students developed to ensure coverage of students who could potentially be experiencing difficult living arrangements. Given the lack of alignment between the teacher and student reporting, the value of collecting information about the living arrangements of school students from a teacher form in addition to the student form should also be considered. Lastly, further development and testing of the survey instruments would be required to reduce reporting error associated with question design.

Any future work to develop a nationally representative survey of school students living arrangements would also be dependent on resources being made available to undertake work on any such collection.

Australian Social Trends – Life after Homelessness, cat. no. 4102.0, March 2012 issue
Census of Population and Housing:Estimating homelessness, 2011, cat. no 2049.0
Information Paper – A Statistical Definition of Homelessness, 2012, cat. no. 4922.0
Information Paper – Methodology for estimating homelessness from the Census of Population and Housing, 2012, cat. no. 2049.0.55.001

This release previously used catalogue number 4111.0.55.001.