Relationships between student mobility and academic and behavioural outcomes in Western Australian public primary schools

The number of times children change schools, or student mobility, is associated with multiple adverse outcomes across the life span. This study used administrative data from the Western Australian Department of Education for public primary school students who completed Year 6 between 2016 and 2019 to examine potential associations between student mobility and academic (using National Assessment Program – Literacy and Numeracy [NAPLAN] participation and scores) and behaviour outcomes (measured through school suspensions). The odds of participating (vs. not participating) in NAPLAN were significantly lower for students with high mobility. High mobility students also achieved significantly lower scores, on average, on NAPLAN literacy and numeracy at Year 3 and Year 5 compared with low mobility students. However, there was no evidence of an association between student mobility and school suspensions. These findings highlight the need for action to address substantial academic detriment for mobile students, many of whom are likely to be from lower socio-economic backgrounds. Furthermore, current policies to address academic disadvantage are likely to exclude those students at substantial academic risk and require revision to be appropriately triaged.


Introduction
Australia is in the top quintile of countries for residential mobility with as many as 40-55% of the population moving residences in any given five-year timeframe (Bell et al., 2017).Residential mobility is strongly associated with movement between schools, particularly for younger students (de la Torre & Gwynne, 2009; Levine Coley & Kull, 2016;Rumberger & Larson, 1998).For families with dependent children, the most important factor in choosing a new housing location is access to services, including schools (Australian Bureau of Statistics, 2013).However, for vulnerable populations, housing instability is the primary motivator for residential changes, and subsequent accommodation and schools are subject to availability rather than choice (Hulse & Saugeres, 2008;Jørgensen & Perry, 2021;Rumberger & Larson, 1998).
The phenomenon of students changing schools, known as student mobility, has been shown in some studies to be related to a number of short-and long-term outcomes for students (Mehana & Reynolds, 2004).Though the reasons for moving schools are wide ranging, changing schools can be associated with reduced academic progress and achievement in the short term (Buckingham et al., 2013;Hattie, 2018;Lu & Rickard, 2016).Hattie (2018) synthesised nearly 1,200 meta-analyses of published research from different countries to identify factors related to student achievement and identified student mobility as one of the top five.An Australian study by Lu and Rickard (2016) found that children who moved schools were less likely to participate in assessments and had poorer academic performance.
Student mobility is an important lifecourse issue.There is evidence that mobile students experience poorer social, emotional and behavioural outcomes (Camacho & Krezmien, 2019;Chen et al., 2011) and have a smaller social network in comparison to their less mobile peers (Bradshaw et al., 2010;Langenkamp, 2016).The ensuing emotional impact of frequent school moves has also been shown to be associated with increased rates of school disengagement, a more negative attitude towards school and declining classroom participation (Gruman et al., 2008).Additionally, there is a strong positive association between mobility and school suspensions in high school aged students (Boon, 2011).In the longer term, greater school mobility has been associated with a greater risk of drug use (Arteaga et al., 2010), violent behaviour (Webb et al., 2016), adult mental health conditions (Mok et al., 2016;Winsper et al., 2016) and increased rates of adult incarceration (Ou & Reynolds, 2010).
In both international and Australian literature, consideration has been given to the demographic characteristics of children who move schools frequently.Mobile students are overwhelmingly from typically disadvantaged families and minority populations (Greenburg et al., 2020;Min, 2021) and are more likely to have parents with lower rates of education who are from low socio-economic backgrounds (Temple & Reynolds, 1999).In Australia, Indigenous 1 students are more likely to move schools frequently in comparison to their non-Indigenous peers (Lu & Rickard, 2016).A number of Australian student mobility studies have been largely centred on communities in Queensland and the Northern Territory with a high proportion of Indigenous students (Danaher, 2012;Doyle & Prout, 2012;Sorin & Iloste, 2006;Taylor, 2012).Prout Quicke and Biddle (2017) suggest that student mobility in Indigenous student populations may be due to cultural incompatibility with Australian education systems.Patterns of Indigenous mobility are also shaped by factors such as cultural connection and kinship (Bourke et al., 2000).
An interesting study population is that of children of military families who move frequently.Here, the data on the relationship between school mobility and student outcomes is inconclusive.There is some evidence that for children of military families, there is little relationship between mobility and academic outcomes for children in their primary school years (McCulloch & Hall, 2016).Some research suggests that children of defence force personnel, who often move schools as their parents are transferred between military postings, may out-perform their mobile non-defence force peers (United Kingdom -Department for Education, 2010).Nonetheless, although children of military personnel who move frequently achieve better academic outcomes than their non-military highly mobile peers, both groups still exhibit lower achievement in English and Mathematics assessments in comparison to their non-mobile peers (Mehana & Reynolds, 2004; United Kingdom -Department for Education, 2010).Any advantage experienced by children of military personnel may be due to Defence Force support structures in place to help with moves (Macdonald & Boon, 2018).
As mobility increases, school attendance tends to become less frequent (Hancock et al., 2013).Highly mobile students also tend to have more unauthorised absences than their less mobile peers (Hancock et al., 2013).
The aim of this research was to determine rates for student mobility in Western Australian public schools, and to explore the relationship between student mobility and students' academic and behavioural outcomes both overall and when stratified by gender and Indigenous status.This study extends the existing literature as it uses longitudinal data for all students in the Western Australian public-school sector, for the entirety of their primary schooling, across four cohorts.This provides a more complete assessment of student mobility and considers the impact of cumulative school moves on outcomes as students advance in grades.Further, we examined the impact of student mobility on behavioural outcomes, for which there is scarce evidence in Australia and internationally.

Data
This was a retrospective observational study with deidentified, longitudinal administrative data sourced from the Western Australian Department of Education through an application to the School Performance Branch in November 2020.It encompassed all students who completed their compulsory, Year 6 primary schooling in a public school between 2016 and 2019.Each student's school enrolment history between Pre-Primary and Year 6 was captured, together with the education and occupation levels for their primary caregivers, student gender, Indigenous status, yearly report grades, Year 3 and Year 5 participation information for the National Assessment Program -Literacy and Numeracy (NAPLAN) and NAPLAN scores.In addition, attendance, coded behaviour and suspension information were available.

Measures for analysis
NAPLAN was introduced in Australia in 2008 and is a large-scale, mandatory standardised test administered by the Australian Curriculum, Assessment and Reporting Authority (Australian Curriculum Assessment and Reporting Authority, 2023).Students in Year 3 and Year 5 who are at school over the NAPLAN assessment period are expected to participate unless they are exempt.Exemptions may occur for students who have arrived in Australia within the twelve months prior to NAPLAN (and do not possess English language proficiency), or for students with significant disabilities (Australian Curriculum Assessment and Reporting Authority, 2016c).NA-PLAN scales are comprised of scores between 0 and 1,000 for each test domain of Reading, Writing, Spelling, Grammar and Punctuation and Numeracy (Australian Curriculum Assessment and Reporting Authority, 2016a).At the time of this study, these scaled scores had been 'constructed so that any given score represents the same level of achievement over time' which allowed comparisons within test domains and also enabled student achievement to be monitored (Australian Curriculum Assessment and Reporting Authority, 2016b).
Several reviews of NAPLAN have identified the advantages and disadvantages of implementing this national standardised test.Amongst the advantages, the Rose review acknowledges the 'recognised benefits' of NAPLAN, including targeting resourcing to the highest need schools and its utility in identifying gaps in educational achievement outcomes (Rose et al., 2020).Additionally, though they concede that many stakeholders have issues with NAPLAN assessments, McGaw et al. (2020), in their NAPLAN Review Final Report, conclude that since 2009, cohort data from NAPLAN assessments has been used to provide information about achievement and assessment participation.Further evaluation of NAPLAN was recently undertaken by the Gonski Institute for Education.The resulting report identified a 'series of issues, and weaknesses, in NAPLAN' including a narrow assessment curriculum and lack of utility in delayed results (Wilson et al., 2021).In terms of NAPLAN participation and scores as a measure for this study, Rose et al. (2020) posit that NAPLAN is best utilised as 'one tool' (p.882) in an assessment toolbox, as it is a snapshot of achievement rather than a more comprehensive view of a child's learning.A number of studies have used NAPLAN participation (and non-participation) and scores to examine a variety of outcomes (Dumuid et al., 2021;Guthridge et al., 2015;Lu & Rickard, 2016;Thompson et al., 2018;Warren & Haisken-DeNew, 2013).Despite its acknowledged limitations, therefore, NAPLAN participation and scores have been used as measures of student achievement for this study.
Student behaviour in Western Australian public schools is recorded on a central system where incident details are accessible by subsequent public schools in which the student enrols, and by relevant Western Australian Department of Education staff.A student may be suspended from school when there is a breach of school discipline causing 'significant disruption to the student, other students or staff' (Western Australian Department of Education, 2021, p. 6).A suspension is considered a serious sanction and is utilised in 'severe circumstances' (Western Australian Department of Education, 2021).Hemphill et al. (2014) found that for students in Grades 5, 7 and 9, higher rates of school mobility increased the likelihood of school suspensions.Two published studies found an association between student mobility and school suspensions in older students (Boon, 2011;Engec, 2006), and it has also been found that the risk of violent offending in adulthood is increased for highly mobile students (Webb et al., 2016).For these reasons, this study has used suspensions as a behaviour outcome to determine if similar associations between school mobility and suspensions are evident in younger students.

Mobility
For the present study, a student mobility measure for each student, based on the number of times a student moved schools after their initial enrolment, was calculated.Students were then categorised as low (0-2 moves), medium (3-5 moves) or high (6+ moves) mobility based on the distribution of this measure.

Academic
We categorised students' participation in NAPLAN in Year 3 and Year 5 according to whether they had participated in at least one of the Reading or Numeracy NAPLAN assessments.All students who had an enrolment record for Year 3 were included in the study.Subsequently, missing data for the Year 5 cohort occurred when students did not have an enrolment record for Year 5 but did for Year 3. Information was not available for the reasons why a student was not enrolled in Year 5, but possible reasons could include when a student had moved interstate or overseas.Following the approach in a number of prior studies (Adams et al., 2020;Brinkman et al., 2013;Laurens et al., 2020), we used the student scaled scores in Reading and Numeracy for those participating in the NAPLAN assessments as a measure of achievement against national standards in literacy and numeracy.

Behaviour
This study used school suspensions as a proxy for student behaviour, categorised as two or more vs.less than two suspensions based on the distribution of this measure.

Statistical methods
To assess the association between student mobility and 1) NAPLAN participation in either Reading or Numeracy, and 2) suspensions from school, binary logistic regression was used.Using only those students who participated in the respective NAPLAN assessments, linear regression models were fitted to assess the relationship between student mobility and total NAPLAN scores (Reading and Numeracy modelled separately).These models were performed separately for Year 3 and Year 5 and were multivariable, adjusting for gender, Indigenous status (Aboriginal and/or Torres Strait Islander), primary caregiver education level and previous NAPLAN information where appropriate.Secondary analyses stratified by gender and Indigenous status were carried out.Model assumptions were thoroughly checked and confirmed to ensure the validity of the analyses and C-statistics assessing model discriminatory power and R-squared values evaluating the proportion of variance explained are also included for the respective models.All analyses were undertaken using SAS software version 9.4, and statistical significance was set at p < .05.

Ethics
Ethics approval was granted through the University of Western Australia Office of Research Enterprise (Human Ethics) on 9 November 2020, Reference 2020/ET000210.

Results
Overall, there were 79,731 individual students completing their compulsory schooling between 2016 and 2019.Demographic characteristics were similar across the four cohorts, where approximately 49% of students were female, 9% were Indigenous and 62% had primary carers who had completed tertiary education.These proportions are consistent with the 2016 Australian Bureau of Statistics census data for gender, Indigenous status and primary carer education levels (ABS, 2020).In terms of student mobility, 93% of students had low mobility, 6% medium mobility and only 1% of students were classified as highly mobile (Supplemental Material 1).
Table 1 presents the students' characteristics stratified by student mobility levels.The distribution of gender was consistent across these groups.Indigenous students were over-represented in the medium and high mobility groups.Students whose primary carer had completed tertiary education had fewer school moves.Most students participated in NAPLAN assessments (98% in Year 3 and 97% in Year 5).In terms of suspensions, most students were either never suspended or suspended only once, with just over 2% of the study population having at least two suspensions.The Year 3 mean (± standard deviation) NAPLAN Reading score for students who participated was 405 ± 93 with mean Numeracy scores slightly lower (389 ± 79).Year 5 students in the study who participated in NAPLAN had a mean score 492 ± 87 for Reading and a mean of 486 ± 72 for Numeracy.Table 2 presents the results from multivariable logistic regression models investigating associations between NAPLAN participation in Year 3 and Year 5 and school suspensions separately.Increased mobility was associated with lower NAPLAN participation rates in both the Year 3 and Year 5 models (all p < .001).Specifically, the odds of not participating in NAPLAN were 3.24 times higher for students in the high mobility category (95% confidence interval [2.68, 3.93]) and higher for students with medium mobility (OR 2.11,95% CI [1.83,2.42]),compared with students with low mobility.Similar results were seen when modelling Year 5 NAPLAN non-participation.
Male and Indigenous students had increased odds of not participating in NAPLAN in both the Year 3 and Year 5 models compared with female or non-Indigenous students, respectively.Students whose primary carer did not finish high school had increased odds of non-participation compared to those who completed tertiary education for both Year 3 and Year 5 models (OR: 1.46, 95% CI [1.31, 1.63]; 1.54, 95% CI [1.37, 1.73] respectively).In addition, those who did not state their education status appeared to have the highest odds of non-participation.In the Year 5 model, those students who did not participate in their Year 3 NAPLAN assessments were more likely to not participate in their Year 5 NAPLAN (OR: 2.89, 95% CI [2.44,3.42]).Male students were more likely than female students to be suspended two or more times (OR: 4.74, 95% CI [4.21,5.35]),as were Indigenous students compared to their non-Indigenous peers (OR: 5.63, 95% CI [5.04,6.29]).Students with primary carers who did not finish high school were more likely to be suspended two or more times compared to students with carers who completed tertiary education (OR: 3.81, 95% CI [3.30,4.40]).Similar to the participation models, those whose primary carer did not state their education level had increased suspension rates relative to students with parents who completed tertiary education.

Analyses stratified by gender and Indigenous status
Subsequently, all models were stratified by gender and Indigenous status separately (Table 4).For both male and female students, highly mobile students had significantly higher odds of nonparticipation in NAPLAN compared to those with low mobility.However, the estimated differences seen were larger for males in the Year 3 model (OR: 3.52 for high vs. low mobility in males vs. 2.94 for high vs. low mobility in females) but were larger for females in the Year 5 model (OR: 3.38 males vs. 4.24 females).In addition, there were larger differences seen between highly mobile and low mobility students for the non-Indigenous students compared to their Indigenous peers in both Year 3 and Year 5 models (Year 3 OR: 4.85 non-Indigenous vs. 3.06 Indigenous; Year 5 OR: 4.66 non-Indigenous vs. 3.71 Indigenous).
The results from the analyses of NAPLAN scores stratified by gender and Indigenous status revealed larger differences on the effect of student mobility across these characteristics in Year 3 relative to Year 5 (Table 5).In Years 3 and 5 Reading models, females with high mobility scored on average 62.4 points and 8.7 points lower than the low mobility females, respectively.When comparing high and low mobility for males, there was an average difference of 62 points for Year 3 and 11.6 points for Year 5 Reading.In contrast, when stratifying by Indigenous status, the magnitude of the difference between the high and low mobility groups for Year 3 was larger in the non-Indigenous students .60 .19 .64 Mean diff = mean difference, CI = confidence interval, N/A = not applicable.

Discussion
The vast majority (93%) of primary school students in Western Australian public schools (2016-2019) either commenced and completed their primary schooling in the same school or moved only once or twice.For school moves, our low category was 0-2 moves, medium was 3-5 and high was 6+.In particular, the low category recognises that there may be good reasons for children to move schools at times, such as military service, enrolling in higher performing schools, changes in residences or families moving for parents' work commitments (Centre for Education Statistics and Evaluation, 2016).However, it is likely that larger number of moves, regardless of the underlying reason, are disruptive to children.Approximately 6% (4,537) and 1.5% (1,178) of study participants  .

65
.62 .49 Mean diff = mean difference, CI = confidence interval.Models adjusted for gender, Indigenous status, primary carer education level and NAPLAN score where appropriate.
were in the medium and high mobility groups, respectively, and these were comparatively small.It is important to note, however, that these percentages represent 5,715 children over a five-year period.Further, these children are at risk of poorer academic outcomes as highlighted through our analyses and other poorer outcomes including social, emotional and longer-term life outcomes as indicated in the international literature (Chen et al., 2011;Herbers et al., 2013;Webb et al., 2016) In our study, Indigenous students and those students whose primary carers did not finish high school had the greatest representation in the medium and high mobility groups (see Supplementary Material 1).As students with these characteristics are already at risk of poor life outcomes (Commonwealth of Australia, 2020; Dickson et al., 2016), increased levels of student mobility have the potential to compound vulnerability, with cumulative effects possible if belonging to more than one of these groups.
As the data used for this study was drawn from administrative data, reasons for school moves were not captured and therefore unable to be examined.Some previous qualitative studies have been better able to determine the motivations behind school moves, especially for Indigenous children in Australia.For example, Prout (2009) found that some Indigenous parents felt compelled to move away from some regional schools due to a perceived 'declining capacity' to cater for their children.Further, some temporary mobility results from sociocultural obligations (Prout, 2009).Researchers in Northern Queensland found antecedents for moving schools included 'family breakdown, cultural issues and problems within the school' (Sorin & Iloste, 2006, p. 234).
In a study from the Australian state of New South Wales, Lu and Rickard (2016) found that student mobility had a negative impact on test participation.Our modelling identified that high and medium mobility students had reduced odds of participation in the national NAPLAN assessments compared with low mobility students.This is critical for two reasons.Firstly, for students who participate, scores from the Australia-wide, standardised NAPLAN assessment provide an academic check-in to identify students as vulnerable in comparison to their peers across the country.Missed NAPLAN assessments are potentially lost opportunities to triangulate other student progress and achievement data.This study found that students who did not participate in the Year 3 NAPLAN assessments were also more likely not participate in the Year 5 NAPLAN assessments.This association indicates that non-participation in the earlier assessments might be a risk factor or predictor for non-participation in later assessments.Consequently, we hypothesise that those students who do not participate in NAPLAN are at increased risk of not being identified as academically vulnerable at school or system levels.Therefore, the opportunity for earlier intervention could be missed, in turn compounding adverse academic outcomes and vulnerability.
Secondly, test participation is critical from the perspective of educational funding allocations as some state-based educational funding is dependent on NAPLAN scores.For example, in schools in New South Wales, a Student Learning Needs Index (SLNI) is determined using NAPLAN literacy and numeracy data from students who have participated (Nous Group, 2018) and forms part of the funding allocation for each school (NSW Government -Education, 2021).Similarly, in Western Australian schools, there is additional resourcing, known as educational adjustment allocation, available to support schools in improving academic outcomes for vulnerable students.For schools, this particular funding allocation gets proportionally greater as the percentage of eligible students in the school increases (Department of Education Western Australia, 2014).Importantly, the eligibility for educational adjustment funding is predicated on the identification of students at educational risk who perform in the bottom 10% of the Year 3 NAPLAN Reading assessment (Nous Group, 2018).Students who do not sit NAPLAN cannot meet this criterion as they have not participated.Hence, there is potentially a policy gap based on our study's main finding that vulnerable students, especially those who move schools frequently, have lower odds of participating in NAPLAN.This suggests that the additional funding may not be being appropriately triaged to the greatest areas of need.Likewise, schools with large numbers of students who do not participate are missing an opportunity for funding which is specifically earmarked for targeted programmes to raise academic outcomes for its students who are at the greatest educational risk.
For the academic outcome of mean NAPLAN scores, this study found that, for those students with increased school moves who did participate in NAPLAN, there was an association with lower mean scores for both Year 3 and Year 5 Reading and Numeracy.This finding is comparable to an investigation into the impact of student mobility on achievement with similar-aged students from inner-London (Strand & Demie, 2006).Our results indicate that the mean difference in NAPLAN scores was, on average, significantly greater for high and low mobility students at Year 3 compared with Year 5, which may suggest a greater need for intervention for highly mobile students prior to Year 3. The magnitude of difference in mean scores across mobility groups was sharply reduced in Year 5 compared to Year 3, though the Year 3 effect differences were still sufficient to determine whether a student's results place them above or below the National Minimum Standard for their year (Australian Curriculum Assessment and Reporting Authority, 2019).
It has been more than a decade since the national Gonski Review of Funding for Schooling found that Western Australia was amongst the states with the lowest funding for vulnerable students (Gonski, 2011).It is difficult to determine how current funding for students at education risk in other Australian states compares to Western Australia as there is no national standard for how students are determined to meet this criterion.Given this study's finding of the association between frequent school moves and the outcome of NAPLAN participation, it can be reasonably argued that student mobility is important to consider when determining funding for vulnerable students in Western Australian schools.This would not be without policy precedent as, in addition to increased rates of school funding for social disadvantage, the education departments in some Australian states presently include school mobility as a measure with associated funding (Graham et al., 2020).
Unlike previous studies (Boon, 2011;Engec, 2006;Webb et al., 2016) which found associations between school mobility and increased rates of school suspensions in older students, our results did not support any association between frequent school moves and increased numbers of suspensions.There are several possible explanations for these findings.Firstly, data indicates that school suspensions are more likely to occur in older students (Graham et al., 2020;NSW Government -Education, 2019) and this study examined the association between school mobility and suspensions in primary school children.Additionally, it is proposed by Green et al. (2019) that school mobility contributes to a lack of school connectedness and that mobile students can struggle to form relationships and 'pull away from peers and adults', leading to increased absenteeism (p. 3-4).Australian research supports the finding that highly mobile students are more likely to have poorer attendance (Hancock et al., 2013).One possible explanation for low rates of suspension in highly mobile Western Australian students may simply be that they are not at school enough to be suspended.It is also feasible that, after a number of school moves, these students are passively disengaged or disconnected to their schooling (Green et al., 2019).As suspensions occur as a result of violent or especially aggressive conduct in schools (Western Australian Department of Education, 2021), it is possible that passively disengaged students do not exhibit the types of behaviours that typically resulting in school suspensions.A further possibility may be that associated demographic factors related to high school mobility mean that vulnerable students are more likely to migrate into lower socio-economic communities, where schools have expertise in supporting behaviours which otherwise might result in school suspensions.It is also plausible that school leaders in schools with highly mobile students, who are the staff responsible for decisions about when to suspend students with inappropriate conduct or behaviour, have an implicit understanding that disengaged students are unlikely to become more engaged in school through school suspensions (Pyne, 2019).

Strengths and limitations
A particular strength of this study was the universal nature of the data set which captured information on all Year 6 students who completed primary school in Western Australia (2016)(2017)(2018)(2019).Specifically, as the data was drawn from the official administrative records from the Western Australian Department of Education, it encompasses all students who undertook primary schooling in the public system in the entire state over this time period.Further, the outcome measures analysed are standardised nationally, allowing for generalisability and comparability.
It should be acknowledged that, notwithstanding the strengths, the study was subject to some limitations.Firstly, the large number of 'Not Stated' for the primary carer's education level made comparisons for this variable difficult.ABS census data includes a category for 'Not Stated' in their reporting of adult education level.For this reason, this category was included in this study and comparisons made accordingly.Secondly, as this study used administrative data, we were not able to determine the reasons for school moves.As motivation for school moves, including factors such as the mobility that accompanies housing instability, or moving as a result of parent employment obligations has been shown to be related to the size and duration of the impact of any association with student outcomes, more qualitative information would have been valuable.Additionally, this study was limited by the inability to determine if the same person was enrolling a student for each school move and therefore, primary carer education level at the time of the child's first school enrolment in Western Australia was used.It is acknowledged that a carer's education level may change over the course of each child's schooling and could be imprecise.There was no capacity, with these data, to identify student mobility to and from private schools or where students transferred into Western Australia from other states or countries.Finally, though our data was extensive and representative of one Australian state, it is acknowledged that differences, such as in geography, vary across Australian states.Therefore, findings from a similar investigation using Australia-wide data in the future would be valuable.

Conclusions and future directions
Public-school funding in some Australian states, including Western Australia, does not incorporate a measure for school mobility or NAPLAN non-participation to calculate additional resourcing for atrisk students.Western Australian public-school funding includes a measure for socio-economic advantage; however, the current study provides evidence that both school mobility and NAPLAN non-participation are measures that should be considered in allocations for students at educational risk.There would be value in further studies considering schools' funding, specifically with reference to students at educational risk in other states and internationally, to propose potential alternatives to the current Western Australian model.
Given this study's findings of the importance of mobility for student academic outcomes, further research is required to investigate the innate diversity in our results.A mixed methods approach may supplement the quantitative findings of studies such as ours with greater detail and would address the limitation of lack of data regarding the reasons for school moves.Furthermore, as a means by which to better understand the movement patterns of mobile students, there would be broad research and policy merit in the development of Australia-wide standards in school enrolment procedures and student mobility tracking.A national approach could support the development of an individual student mobility measure which would add value beyond what is currently available and would include interstate, as well as intrastate, school moves.In addition, Prout and Hill (2012) suggest that one goal for future research in studies of Indigenous student mobility would be for better school enrolment measures to ascertain source and destination schools.Though it is acknowledged this would not be 'sufficient to fully understand the relationships between Indigenous mobility and formal schooling' (p.65), it would contribute to a required evidence base for educators and those in policy formation to effectively resource support services for mobile students (Prout & Hill, 2012).
A final recommendation from this study is that all staff and schools, especially those with higher mobility than average, be encouraged to implement policies and practices that welcome, engage and further support transferring students on their enrolment in a new school.While it could be assumed that such schools have informal mechanisms and structures in place for welcoming and engaging students who are new to the school, there would be utility in the development of more formal processes to support highly mobile students.A number of successful initiatives, responsive to the needs of particular communities, include a Queensland pilot study where Mobility Support Teachers were introduced to assist highly mobile students with their learning and to facilitate connections to the school community (Hill et al., 2011).A similar program, The School-Based Attendance Officer Program, has operated in the Kimberley region of Western Australian (Prout & Yap, 2012).Further, a small case study in a rural northern Queensland community with a 'reputation for being skilled in catering for mobile students' (p.20) were intentional in working to include families in their school (Henderson, 2017).Staff were deliberate in their interactions to ensure parents were made to feel welcome, supporting the notion that initiatives responding to mobile students need to go beyond the students themselves.In addition to whole school responses to this phenomenon, individual classroom teachers are critical in supporting mobile students.The work of Butt et al. (2016) recognises several practices broadly categorised under the headings of 'Creating a welcoming environment', 'Establishing initial social support', 'Checking where a new student's learning is at' and 'Engaging students in literacy learning' to engage students who enrol at a new school.While it is certainly the case that initiatives to support mobile students should be context and community specific, these studies outline the possibilities and underscore the benefits to both individual students and also to schools with targeted programmes for incoming children.
Further research on the potentially mitigating impact of such practices would be beneficial, particularly for schools with large proportions of mobile students.
a counts and percentages (n (%)).b for those who participated in respective test (mean ± standard deviation).c Students had a score for Year 3 NAPLAN but not for Year 5.

Table 1 .
Baseline summaries of the study population by student mobility group.
Notes: Cohort summaries contained in Supplementary Material.

Table 3 .
Multivariable linear regression results modelling Year 3 and Year 5 Reading and Numeracy NAPLAN scale scores.

Table 4 .
Stratified multivariable logistic regression results displaying student mobility group comparisons for Year 3 and Year 5 NAPLAN participation, Year 5 NAPLAN participation and suspensions by gender and Indigenous status.
1Modelling the outcome no versus yes. 2 Modelling the outcome 2+ versus 0-1; models adjusted for gender, Indigenous status, primary carer education level and Year 3 participation where appropriate.

Table 5 .
Stratified multivariable linear regression results displaying student mobility group comparisons for Year 3 and Year 5 NAPLAN Reading and Numeracy scale scores, by gender and Indigenous status.