British Educational Research Journal

Introduction

Achieving high school credentials is recognised internationally as a critical factor in the successful transition to adulthood. It constitutes a key platform upon which financial security, individual and societal wellbeing are based (Lucas, 2001; Howieson & Iannelli, 2008; Frønes, 2010). There is a strong association between failure to achieve educational credentials and later disadvantage (Strathdee & Engler, 2012). Considering the Scottish context, Howieson and Iannelli (2008) note that high school credentials are important predictors of potential future income, and having high school qualifications reduces labour market marginalisation. While New Zealand performs well in international league tables for overall educational achievement, there is a significant minority of children who fail to achieve school qualifications and who thus face increased social and economic risks across the life course (Advisory Group on Conduct Problems, 2013). Understanding the factors that precipitate premature disengagement from school is clearly important.

The risks of high school non‐completion are not evenly distributed in the population. Furthermore, non‐completion does not always directly relate to academic ability and cannot be blamed solely on the individual (Strathdee & Engler, 2012; Becker & Tuppat, 2013). Indeed, factors in the lives of individual youths, within their families and communities, as well as factors within schools themselves, are all implicated in the capacity of youths to successfully graduate from high school (Ungar & Liebenberg, 2013). The central role that educational institutions play in marginalising young people has been noted elsewhere (Parffrey, 1994; Frønes, 2010): ‘as education became the mandatory road to integration, educational institutions also came to constitute a fundamental mechanism of marginalisation’ (Frønes, 2010, p. 314).

Howieson and Iannelli (2008) argue that schools are a major site where class inequalities are reproduced. Of particular relevance to the current study, they note that even minimal school qualifications are important for vulnerable youths that can tip life chances in the direction of more favourable outcomes. For disadvantaged youths, being able to stay at school appears to be the best way to achieve educational qualifications and from there to improve their prospects (Howieson & Iannelli, 2008).

Because of the powerful role education plays in adult outcomes, there is considerable interest in the factors that influence the capacity of young people to accumulate educational credentials. Much of this research has focused on individual‐level factors that are associated with premature exit. For instance, recent research has identified that prior grades are the strongest predictor of GPA (grade point average) and that psychosocial and behavioural factors add incremental validity to GPA predictions at high school (Casillas et al., 2012). Martin (2011) found that repeating a year of schooling had a negative influence on a range of individual factors all of which had implications for gaining high school qualifications. Mental health issues have a well recognised relationship with lower levels of educational attainment (Kessler et al., 1995) and of all childhood and adolescent‐onset psychiatric disorders, only conduct and attention disorders (ADHD) explain reduced rates of high school graduation (Breslau et al., 2011). Engagement in the youth justice system is also associated with increased risks of premature school disengagement (Lochner & Moretti, 2004; Bottrell et al., 2010; Prior & Mason, 2010).

School engagement is a relatively new field of research that seeks to understand the factors that propel young people out of mainstream schooling (Frawley et al., 2014). High levels of engagement predict achievement of tertiary qualifications and also translate into higher status occupations (Abbott‐Chapman et al., 2014). Family and school characteristics also influence educational engagement even when background levels of risk are controlled for (Christle et al., 2007; Sharkey et al., 2008). For instance, teacher–pupil relationships predict engagement and performance (Furrer & Skinner, 2003). Relationships with parents also influence student educational motivation (Cheung & Pomerantz, 2012). Factors within the control of schools themselves such as school climate (Christle et al., 2007), collaboration between schools and parents and between teachers and students (Ravet, 2007), student locus of control and involvement in school activities (Finn & Rock, 1997) all increase levels of student engagement regardless of student background.

It is increasingly recognised that the interactions between a broad range of factors from the micro‐level to the macro‐systemic influence school graduation rates by vulnerable youths rather than one or two factors in isolation (Fredricks et al., 2004; Martin, 2011). Factors beyond the control of individual students have been found to play a role in levels of engagement (Rumberger & Thomas, 2000). Fernandez (2008) reports on outcomes for youths in the welfare care system, and notes that foster placement stability appears to improve outcomes for youths. She also interestingly highlights the reverse pattern where stability and predictability in the school environment contributes wellbeing benefits to children in foster care.

A limitation of much of the research conducted to date is that it has been located within schools. This means that our understanding of disengagement has been shaped predominantly by the experiences of students who are actually attending school. These studies thus are somewhat limited in their capacity to tell us about the experiences of youths who face high levels of adversity, because those facing the most challenge typically disengage early and remain outside the education system (Sodha & Guglielmi, 2009; Ungar & Liebenberg, 2013). Those studies that have drawn community samples conclude that different protective factors play a role in engagement for youths who face high levels of background risk and adversity compared with those not so exposed. For instance, in their community‐based study of engagement Cooper and Crosnoe (2007) reported that while high parental involvement was a protective factor for disadvantaged youths, it was not a protective factor for high socio‐economic status youths.

While a number of students remove themselves prematurely from school, others are excluded by schools. Certain groups of youths are more likely to be excluded, and those from deprived backgrounds, children with special educational needs and those in the child welfare care system are disproportionately represented in the population of students that schools expel (Sodha & Guglielmi, 2009). It has been suggested that efforts to promote active citizenship and participation at school have the potential to draw disengaged youths back into school because they take account of student perspectives and the meanings they attach to school involvement (McCluskey et al., 2013).

The analysis presented here seeks to advance debates about young people's education experiences by examining the interactions between a range of individual, relational, school and service‐related factors, and the capacity of a group of of young people exposed to high levels of adversity to stay on‐track with their education. On‐track is defined as young people maintaining themselves at about their age‐level in terms of achievement of educational credentials and enrolment in the year‐level appropriate to their age (see Methods and Measures). Despite confronting atypically large amounts of risk and adversity, some of the young people managed to achieve academically, however, most did not. Understanding how to best support such youths to sustain academic progress despite risk and adversity is important. What can we learn from considering the experiences of vulnerable youths as they move through adolescence? How do they manage to sustain educational involvement and to achieve at approximately their age‐level? Do the additional educational supports put around them help them to stay on‐track? Do other services help them to continue with their education? What role do their own characteristics, their friendships, their families and communities play in helping them to achieve academically? It is to be expected that given the complex and fluctuating nature of these youths’ lives, no single factor will explain how they stay on‐track with their education. Accordingly path analysis was used to model the interactions of a number of key variables with the extent to which youths were able to stay on‐track. We hypothesised that (see Figure 1):

Hypothesis: The mix of risks and resources that youths facing adversity bring into educational contexts have an impact on their capacity to stay on‐track with their education. This impact is mediated by the practices of schools and by the provision of formal supports and services.

Specifically:

• H1 There is a positive relationship between the resource independent variables (positive school environment, parent/caregiver relationship, positive peer group, social participation).
• H2 There is an inverse relationship between the resource and risk independent variables (individual externalising, individual internalising, exclusion from mainstream school).
• H3 There is a positive relationship between the risk independent variables.
• H4 Youths who score higher on resources will score higher on the on‐track measure than youths who score lower on these resource measures.
• H5 Youths who score higher on the risk measures will score lower on the on‐track measure than youths who report lower levels of risks.
• H6 Formal services outside of school (mental health, youth justice, child welfare) will mediate the combined impact of risks and resources on these youths' capacities to stay on‐track with their education.
• H7 Participation in mainstream education as well as formal educational supports provided inside schools (additional to mainstream classrooms) and delivered in the community (such as alternative education programmes), will mediate the combined impact of risks and resources on youths' capacities to stay on‐track with their education.

Methods

Results

Two changes were made to the original model based on modification indices; the exclusion variable was changed from an independent to a mediator variable, and child welfare and youth justice services were removed as they did not add any unique explanatory power to the model. These changes resulted in a model with excellent fit [χ² (26, N = 605) = 20.816, p = 0.751, CFI =1; GFI = 0.994, AGFI = 0.984, RMSEA = 0.000].¹

The path analysis produces correlations between a range of independent variables (shown by curved double‐headed arrows at the top of Figure 2) and it also calculates regressions between these variables and the mediator variables (shown as the boxes and the single‐headed arrows in the middle of Figure 2), it then demonstrates the amount of variance all these relationships account for in the dependent variable (shown as the single box at the bottom of Figure 2). In terms of the relationships between variables as proposed in the seven hypotheses, the following patterns were observed in the data (see Figure 2).

As expected, the resource‐related independent variables (the resource variables at the top of Figure 2) all showed positive relationships with each other. However, contrary to expectations, social participation was only positively related to school environment, and the relationship between the positive peer group and parent/caregiver relationship variables was not significant. As a general principle, however, if youths had high scores on one resource measure they had high scores on all the other resource measures. There was thus partial support for the first hypothesis.

Hypothesis 2 proposed that as scores on the resource measures increased, scores on risk measures would decrease (that is; the four resource variables at the top of Figure 2 would decrease as the two risk variables increased). There was partial support for this hypothesis. Specifically, internalising and externalising risks had significant inverse relationships with both the positive school environment and parent/caregiver variables. Externalising risk also had a significant inverse relationship with the positive peer group variable (r  =  –0.651, n = 605, p ≤ 0.001).The two risk variables also had a positive relationship with each other. Externalising risk had a strong positive correlation with school exclusion mediator (β = 0.470, n = 605, p ≤ 0.001) and a small positive correlation with internalising risk (r = 0.073, n = 605, p = 0.019). However, there was a significant inverse relationship between internalising risk and exclusion (β = –0.105, n = 605, p = 0.003). The third hypothesis thus also finds partial support in the data.

In terms of hypothesis 4, of the five original independent variables, only social participation had a direct relationship with the on‐track variable. Paradoxically, the parent/caregiver variable had a small direct relationship with exclusion, that is; stronger relationships with parents/caregivers were linked to an increased likelihood of school exclusion. Neither a positive peer group nor a positive school environment had a direct impact on capacity to stay on‐track, but both variables had a positive effect on capacity to stay in mainstream education, and this, in turn, had the strongest impact on capacity to stay on‐track. These patterns provide only partial support for hypothesis 4.

Initially, school exclusion was proposed as an independent variable. However, the path analysis indicated that it functioned as a mediator rather than an independent variable and it was moved accordingly in the final model. Externalising and internalising risks did not have a relationship with being on‐track; rather they influenced levels of exclusion from school, participation in mainstream classrooms and the amount of service received from mental health and additional educational services. Mental health service use had a positive impact on capacity to stay on‐track and, in this way, for youths with high externalising and/or internalising risks mental health services functioned as a pathway to staying on‐track. However, there was another, less successful pathway for youths with high levels of externalising and internalising risks and this was through school exclusion and involvement in additional educational services (such as resource teachers in schools, or community‐based programmes). Neither of these pathways increased the chances that youths would stay on‐track and both school exclusion and additional educational services significantly increased the chances that youths would not return to mainstream classrooms. The data find only partial support for hypothesis 5.

Hypothesis 6 concerned the impact of formal services outside of the education system (child welfare, youth justice, education additional to mainstream programming, mental health) on the capacity of youths to stay on‐track. Only mental health services directly facilitated youths' capacity to stay on‐track. Neither involvement with child welfare nor justice services had a significant impact on the capacity of youths to stay on‐track. Removal of these two variables from the analysis improved the fit statistics for the model. In the final model, the resource‐related factors within youth environments had a greater influence on their capacity to stay on‐track than did formal services. In particular, social participation appeared to play a direct role in assisting this group of vulnerable youths to stay on‐track with their education. The analysis shows partial support for hypothesis 6.

Of educational services, only attendance at mainstream schools had a positive influence upon on‐track status. Mainstream schools exerted the single largest influence on youths' ability to stay on‐track; other formal resources provided through the education system (either within mainstream schools or through community programmes) appeared to have no impact on the capacity of this group of vulnerable youths to stay on‐track. The data provide only partial support for hypothesis 7.

The partial support for all of the hypotheses and the modifications to the original model indicated that the pathway to staying on‐track with education was complicated for this group of vulnerable youths.

Discussion

Staying on‐track with education is not a simple matter for vulnerable youths. The model illustrates complex interactions between the characteristics of the social environments (risks and resources) around vulnerable youths and the formal services designed to assist them. These factors combine to shape their capacity to progress their education, sometimes in paradoxical or counter‐intuitive ways.

Resources and risks as pathways to educational achievement

Of all the independent variables, only social participation directly impacted upon youths on‐track status. Others have found that social participation can play a positive role in drawing disengaged youths back into mainstream classrooms (McCluskey et al., 2013). In the current study, social participation did not function in this way. However, it did increase capacity to stay on‐track independently of mainstream classroom participation. Community resources then appear to contain resources that could be utilised when supporting vulnerable students to continue with their education (Sanders et al., 2013).

A positive parent/caregiver relationship was linked to increased rates of exclusion from school. On the face of it, this relationship appears to contradict established research that shows strong parent–youth relationships positively influence educational involvement (Cooper & Crosnoe, 2007; Cheung & Pomerantz, 2012). However, it does not necessarily follow that because positive parent–youth relationships predict educational engagement that students who are excluded will necessarily have poor relationships with their parents/caregivers (McMullen, 2004; Liebenberg et al., 2015). Children having difficulties at school can also have strong bonds with parents. Indeed, elsewhere it has been observed that high risk youths are not a homogenous group (Foster & Spencer, 2011).

Even among high risk youths, such as those in this study, it may well be that those who manage to stay engaged in education are different from those who disengage, and further, among the group of youths who are excluded there may be critical differences on dimensions such as relationships with parents/caregivers. Support for this argument can be found in an allied body of literature concerned with truancy. Here, complex intersections between pupil, school, parental and community factors have been observed to shape patterns of student disengagement and further, while teachers consistently attribute student disengagement to failings on the part of parents, parents and pupils consistently attribute this to failings in school (Malcolm et al., 2003). Regardless of the explanation, in the current study 8% of the excluded young people reported strong parent/caregiver relationships and this highlights the presence of positive resources in the youths' social ecologies that were potentially available to schools and other professionals to assist them with re‐engagement.

Experiencing school as a positive place and having a positive peer group both made significant contributions to attendance at mainstream schools (which was the largest contributor to staying on‐track). Elsewhere the importance of strong positive school relationships and engagement practices in supporting vulnerable youths through adolescence and in keeping them achieving educationally have been noted (Furrer & Skinner, 2003; Prior & Mason, 2010; Frawley et al., 2014; Sanders & Munford, 2015). It is particularly notable here that the two measures that captured relationships that occur at school were such strong predictors of mainstream attendance while additional educational supports, those services provided precisely to assist vulnerable youths to stay at or return to school, reduced the likelihood of mainstream participation.

The impact of mental health issues upon the capacity of youths to continue with their education has been extensively canvassed in the literature (see for example, Kessler et al., 1995; Lochner & Moretti, 2004). In the current study, youths scoring high on the internalising risk measure were less likely to be excluded from school and they were also more likely to continue participating in mainstream education and to use mental health services than youths who had low scores on this measure. This is consistent with findings elsewhere (Breslau et al., 2011) and in the qualitative findings this pattern was also observed (Sanders & Munford, in press). Externalising risks had a strong relationship with exclusion from school, and also with levels of additional educational and mental health service utilisation. Two different pathways were observed depending on the type of individual risks youths brought into schools. These patterns suggest that of the psychological, emotional, learning and behavioural challenges children bring into schools those that do not generate overt behavioural challenges for schools (i.e. depression) are most likely to be accommodated (Advisory Group on Conduct Problems, 2013). Young people with high levels of externalising risk were most likely to be excluded and smaller numbers were picked up by the mental health system. When young people did attend mental health services these services provided an avenue for staying on‐track. However, almost half of those youths reporting high levels of externalising risks were excluded from school and while some received additional educational supports neither this, nor exclusion, increased the chances that they would stay on‐track with their education. The way in which the education system responds to externalised risk behaviours is clearly a critical ingredient in continuing with education. In the early 1990s Parffrey (1994, p. 117) commented on the reluctance of schools to persevere with students with high levels of behavioural needs or risks and she suggested that there was: ‘[a] systemic abdication of responsibility. Certain children, it would seem, are not wanted. They do not fit, behaviourally, socially or emotionally. Schools successfully get rid of them.’ While the data from the current study suggest that schools and mental health services appear able to facilitate continued integration of youths with depression (internalising risk), the same cannot be said for youths who bring other types of risk with them to school. Rather than equalising wider disadvantage for these youths, schools appear to reinforce it (Howieson & Iannelli, 2008). Others have also remarked upon the ways in which system behaviours open up pathways to different sorts of outcomes with longer term consequences for young people, entrenching patterns of disadvantage in the process (Parffrey, 1994). For instance, Bottrell and colleagues (2010, p. 67) note that the apparently benign bureaucratic procedures that surround school exclusion and the time delays they involve, ‘leave students disconnected from school [and] are for some young people pathways between educational and youth justice systems’.

Educational services as pathways to educational achievement

The path analysis clearly highlighted the critical role that being able to stay at mainstream school had upon educational achievement for vulnerable youths, most of whom had experienced disruption for at least a part of their school careers. The final model suggests that when young people require access to additional educational support and when they are excluded from regular classrooms it is unlikely that they will return to a mainstream school. Moreover, the likelihood that they will achieve school qualifications diminishes significantly. This pattern points to a major role for mainstream schools in working with the most vulnerable students (the students who are most likely to either remove themselves or be excluded) to keep them engaged and attending mainstream classes. This is consistent with work elsewhere, such as that of Howieson and Iannelli (2008) who found that being able to stay at school was critical for educational success of marginalised youths. Youths with high utilisation of additional educational services were, paradoxically, less likely to stay engaged at mainstream school, despite the fact that a number of the services included were actually provided on school premises (such as resource teachers). These services did not appear to be able to successfully re‐open pathways into mainstream classrooms or to increase the chances that young people would stay on‐track. In 1994 Parffrey (p. 107), commenting on the situation in the UK, predicted just such a result; that increased competition between schools would result in vulnerable children not being accepted back into mainstream classrooms after receiving support from non‐classroom‐based programmes, such as, in the current study, alternative education programmes. She noted that in the reforms of the 1990s ‘pastoral care and the teaching of children with differences—particularly behavioural and emotional difficulties—[was] but one of the[se] curriculum casualties’. The current study suggests that the quality of the school environment, as it was experienced by youths, facilitated retention in mainstream classrooms in ways that additional educational services did not. It may be then, that retaining vulnerable youths at mainstream schools is not so much an outcome of the additional programmes offered by the education system as it is a result of the quality of the relationships teachers and other professionals in the school environment build with these students while they are still attending school. These are factors over which schools do have direct control and from this analysis they include ensuring that they provide a welcoming and safe environment for vulnerable children and encouraging positive peer interactions (Furrer & Skinner, 2003; Frawley et al., 2014). The data from this study appears therefore to support the conclusions drawn elsewhere (see for example, Howieson & Iannelli, 2008; Frønes, 2010) that the contemporary education system functions to crystallise existing inequalities rather than to compensate for other disadvantages experienced by marginalised young people.

Other formal services as pathways to educational achievement

The original model proposed that a range of services (child welfare, youth justice, education programmes additional to mainstream classrooms and mental health) would facilitate re‐engagement of youths in education and support them to stay on‐track with gaining qualifications. The analysis indicated that only mental health services made a unique contribution to these youths' capacity to stay on‐track. None of the other formal services assisted youths to stay in mainstream education (which was the strongest predictor of staying on‐track). Interestingly, youths who reported school was a positive place were both less likely to use mental health services and more likely to stay in mainstream classrooms, suggesting a potential buffering effect that positive school environments have concerning youths' mental health (Bennett & Coggan, 1999).

Formal services did not compensate for the things that were missing in the environments around vulnerable youths. Rather, the resources around youths, notably positive peers and levels of social participation, appeared to play a more significant role in keeping youths in mainstream education and on‐track respectively, than did formal services designed to assist young people facing high levels of adversity.

While mental health professionals exerted a small but nonetheless positive influence on the capacity of vulnerable youths to stay on‐track, it is of concern that services provided through the child welfare and the youth justice systems did not play such a role for this group of youths. It has been observed elsewhere (Sodha & Guglielmi, 2009) that engagement in these systems is associated with increased risk of early disengagement from school and reduced chances of obtaining school credentials (Bottrell et al., 2010). This is not to say that these systems necessarily create this risk. Rather, the lack of a positive connection between these services’ involvement in the lives of vulnerable and marginalised youths and improved chances of educational success is of concern.

Limitations

This study is cross‐sectional in its design and, ideally, it needs to be supplemented by future research that includes longitudinal data collection strategies, including observing how systems facilitate—and fail to facilitate—educational outcomes for vulnerable youths (Murray & Zvoch, 2011; Martin, 2013). Also, as the study takes the unconventional step of sampling youths who have already used at least two formal services, the sample offers a unique glimpse into the influence that multiple services have upon educational outcomes given the presence of a range of risks and resources, but at the same time this method poses some challenges to the external validity of the findings. It is not known whether young people who use multiple services are a special population whose deficits and challenges influence these patterns. This, therefore, limits the generalisability of the findings. Furthermore, sampling only youths who are multiple service users stands to introduce selection bias to the study. While the purpose of selecting multiple service users was to ensure a sample of youths facing some of the highest risks in their communities, there is still a chance that those youths who are the most vulnerable are not in fact receiving services and are therefore not represented in this study. The data are also self‐reported and reflect student accounts only. Gaining insight into teachers’ perspectives (as in Murray & Zvoch, 2011) of the ways in which schools intersect with other service systems, draw upon the positive resources around youths, and address internalising and externalising risks, would deepen the understanding on how to improve the chances of vulnerable youths can best be supported to stay in school.

Conclusion

Keeping young people on‐track when they face a significant amount of adversity requires the combined efforts of formal services as well as the resources present in young people's own lives. Efforts need first to be directed at keeping these youths attending mainstream schools as this was the strongest predictor of young people staying on‐track. However, as could be seen from the analysis, a large proportion of these youths either exclude themselves, have been excluded, or face significant obstacles beyond their control that stop them from regularly attending mainstream schools. Therefore, directing attention at how to make school a positive environment for vulnerable children and how to facilitate or support their ongoing attendance is clearly very important, particularly given the fact that most of the young people interviewed placed a high value on gaining an education. Of particular concern is the lack of a direct contribution that additional educational services made to positive educational outcomes for these youngs people; these services did not appear to play a positive role in either reintegrating them into mainstream settings or with keeping them on‐track. The model suggests that the way in which mental health services work with vulnerable young people may hold some clues as to the components of effective practices in terms of educational engagement as there was a direct relationship between mental health service use and youths' capacity to stay on‐track. The model also points to a need to consider alliances between welfare/justice programmes and educational services both within and outside of schools because neither of these service systems made a unique contribution to youths' capacity to either stay on‐track or to stay at school. Educational engagement is a concern for those working in both of these systems because educational progression has such a strong relationship with better outcomes. Fernandez (2008) has demonstrated that integration between services and educational providers produces benefits for young people in terms of enhancing both care and educational outcomes. Finally, the model points to the importance of taking account of the interaction between youths' backgrounds—their friends, their parents and involvement in community activities—and other factors, such as service provision, in terms of enhancing their capacity to stay engaged and achieving in education. The model suggests that there are positive resources around youth that can be harnessed to produce educational gains for the most vulnerable young people in our communities, but that not every positive resource produces a positive educational gain. For this reason, it is important that educational and other professionals carefully examine these domains (friends, family and community) to identify which of these has the potential to assist with educational engagement and then intentionally work to maximise the benefits for vulnerable youths. Without this, the most vulnerable youth in our communities will continue to experience their involvement in formal services as gatekeeping rather than as facilitating advancement and creating opportunity.

Acknowledgement

The authors thank all the young people who participated in this study. They also acknowledge the contribution of The Donald Beasley Institute, the Victoria University Research Trust and its staff, Youthline Auckland, KapitiYouth Support and Otago University and, importantly, the Ministry of Business Innovation and Employment for funding the research.