Introduction
Fixed period and permanent exclusions are used in schools in the United Kingdom as a method of tackling the more severe forms of student misbehavior, such as physical violence, or persistent disruptive behavior. Exclusion (also known as suspension in other jurisdictions such as the United States) is used to remove disruptive students from classrooms on a temporary or permanent basis. However, research suggests that exclusions are associated with poor academic and occupational outcomes, externalizing behavior (such as criminal activity), and internalizing behavior problems (such as self-harm; e.g., Gazeley et al.
2013; Lanskey
2015). High proportions of juvenile offenders and prisoners report having been excluded from school prior to being convicted, suggesting that exclusion is situated somewhere on a trajectory to later offending and incarceration for many students (e.g., Challen and Walton
2004). Furthermore, a recent study by Perry and Morris (
2014) found that students attending schools that exclude more frequently than other schools appear to suffer academically,
whether or not those students are excluded. This is in contradiction to an often cited justification for exclusion as a policy, namely that the disruption caused to other children is unfair and risks their educational achievement (e.g., Noguera
2003; Perry and Morris
2014).
Punishment for misbehavior at school is the first time that many children are sanctioned outside of the home. How this is carried out by the school and perceived by the child may have important consequences for later life, but more immediately for their socio-emotional development and educational attainment. Students who are excluded may show escalations in the negative behaviors that led to the exclusion if they perceive this sanction as unfair (e.g., Piquero et al.
2004), feel stigmatized by being excluded and/or feel no, or deny feeling, shame about being excluded. In addition, by being labelled as a “bad guy”, students may identify themselves with this label and through a self-fulfilling prophecy (Bernburg and Krohn
2003) engage (more) in the behaviors that originally led to this label. Further, by being excluded from school, adolescents may also have more opportunities to spend time in environments conducive to crime (e.g., Wikström et al.
2012). Exclusion from school is also the most explicit form of rejection by the educational system (Munn and Lloyd
2005). Therefore, there is also a risk that exclusion could weaken students’ perhaps already fragile relationships and engagement (bond) with school, through removing the fear of punishment and/or making them feel rejected. Either way, exclusion signals that further help may be needed by the student and/or the school.
What also calls into question the defensibility of relying on exclusion as a sanction for misbehavior is that, in the case of fixed-period exclusions, students in England and Wales have few demands placed on them while excluded, and receive minimal support upon returning to school. Schools are required to set and mark work for exclusions lasting more than one day but are only required to arrange alternative education after the fifth day of a fixed-period exclusion. While guidelines require schools to have a strategy for the reintegration of students upon return to school after a fixed period exclusion, there is no further clarification on what this should constitute. Moreover, there are no mechanisms in place to check the degree to which these guidelines are followed (Department for education; DfE
2012). For policy-researchers, this means that the deep irony of exclusion as a “punishment” is that for some children who are not bonded to school, exclusion is viewed as nothing more than a school sanctioned “holiday” (Dupper et al.
2009).
Children and adolescents at the highest risk of school exclusion experience a variety of vulnerabilities, including mental health problems, learning difficulties, experiences of maltreatment in and outside of the home, poverty, and other risk factors. Students who are excluded tend to be “hard to reach”, disruptive and in many cases aggressive toward adults and/or other peers. Exclusions are also not meted out to all students equally. Over-represented groups include male students, students from low socio-economic groups, students with special educational needs, and ethnic minorities (e.g., Gazeley et al.
2013; Office of the Children’s Commissioner (OCC)
2012; OCC
2013). Those excluded may not like school in the first place, perhaps partly as a result of finding school difficult due to their (unmet) educational needs (DfE
2012). Moreover, while official records are kept for permanent exclusions, fixed-period exclusions in the UK have been less systematically monitored or entirely unrecorded at times (Osler and Hill
1999), leading to underestimates in the numbers of exclusions. Furthermore, the issue of “illegal” and unrecorded exclusions complicate attempts to understand the full impact of exclusions (OCC
2013).
In summary, exclusion is widely used in the UK, but evidence suggests that it is an ineffective—and even potentially harmful—way of dealing with students with problematic behavior (Gazeley
2010; Osler and Vincent
2003). While interventions targeting behavior problems and school exclusion in youth exist and are implemented in many schools, few of them have been subjected to a rigorous evaluation. It is therefore not clear if and to what extent they are effective. For this reason, in the current study we evaluated a pre-existing intervention that aimed to decrease school exclusions and related problem behaviors.
The Intervention
To procure an intervention for this evaluation, the research team approached the Education Endowment Foundation (EEF), which specialises in funding randomized controlled trials (RCTs) in school settings. A bidding process organised by the EEF sought to identify a suitable program and provider and drew up a shortlist of potential interventions. Catch22 and its Engage in Education—London (EiE-L) program was selected for evaluation as it had the clearest description of aims and mechanisms of change, and also presented promising findings from a preliminary evaluation (see Catch22
2013). In addition, Catch22 has delivered a range of programs to high-risk youth of varying ages in the UK.
The EiE-L intervention aimed to improve students’ behavior by developing their communication and broader social skills. EiE-L operated at the individual, school and family level. It aimed to provide targeted support to students with issues they were particularly struggling with, support teachers in addressing the behavioral and communication needs of students, and assist families to better support their children in school. The program consisted of one-hour long group and one-to-one sessions with students over 12 weeks. Each group session was delivered by two core-workers who were assigned to a school and one-to-one sessions were delivered by one of these core-workers. Based on prior experience working with youths, twelve core-workers were recruited and employed on a one year contract specifically to deliver this intervention. Core workers attended a four-week long training program on the principles and delivery of EiE-L. Material for group sessions was developed in conjunction with I CAN, a specialist communication charity, and addressed different aspects of communication difficulties (understanding, language processing, expressive language, and social communication) as well as social and behavioral issues. Session content and the resources required for delivering each session (e.g., scheme of work, session plans, session worksheets) were described in a guidebook available to each core worker at the time of the training. Sessions focused on interpersonal social skills such as effective anger management skills, assertive communication skills, or learning to appreciate the availability of different response alternatives in a variety of situations. See Table
1 for the description of the curriculum and main goals of each of the 12-sessions. One-to-one sessions were used to build on themes covered in group sessions or help participants with specific problems at home or at school. Home visits or phone calls to parents allowed the intervention providers to maintain engagement in the intervention by informing parents or carers about the performance of their child, positive or negative. Finally, I CAN provided support for teaching staff by delivering training sessions, conducting observations and conducting additional follow-up sessions. Please see the study protocol (Obsuth et al.
2014) for a full description of the intervention.
Table 1
EiE-L session goals and description
1. The skills I start with | To learn effective communication skills. Participants are invited to think about their strengths and difficulties in regard to their communication strategies with teachers and peers |
2. Managing difficult emotions | To learn effective anger management skills. Participants are made aware of a range of emotions, the triggers for some emotions and some alternatives for managing them |
3. Understanding conflicts | To learn strategies for self-calming and de-escalating confrontations |
4. I have choices | To learn to appreciate the availability of different alternatives in a range of situations, to appreciate choices; their causes and effects |
5. Check it out | To learn to identify difficulties in comprehension; being aware of confusion by instructions; positive skills and attitudes to ask for extra explanations (e.g., interrupting appropriately) |
6. Different talk for different people | To learn to adjust the way of talking depending on one’s conversation partner and location. Develop an understanding of the difference between formal and informal communication exchanges |
7. Looking back looking forward | Evaluate personal performance and setting goals for the second part of the course |
8. Co-operating with others | To learn assertive communication skills in-group situations. Discussing with others in small groups, accepting others’ opinion, changing personal opinions |
9. Aggressive, assertive, passive | To learn to understand and be aware of different styles of communication (aggressive, assertive, passive) and develop skills for adaptive, assertive interchange |
10. Communication without talk | To learn to understand body language and non-verbal signals. To be aware of potential biases based on non-verbal signs/stereotypes (dress, ethnicity, posture, etc.) |
11. I can change my world | To learn to identify and acknowledge personal difficulties with classroom behavior and identify strategies to improve |
12. Summing up | Final session summarizing the learning process, relevance of communication skills, personal achievements and personal challenges |
Theory and Research Supporting the Intervention
The focus on the social aspects of communication and broader social skills represented the theory of change endorsed by the intervention provider. This theory of change appeared plausible in the context of other research suggesting that students who are excluded often have social-skills and social communication difficulties which may compromise their ability to benefit from the curriculum and behave prosocially (Clegg et al.
2009). Links between social, cognitive and interpersonal communication difficulties and behavioral problems at school have been identified in the literature. Researchers suggest that social-cognitive processes such as social communication problems (e.g., Gilmour et al.
2004; Moffitt and Scott
2009), social-emotional learning difficulties (Durlak et al.
2011), agency skills (e.g., Larsen and Angus
2011) and deficient social competence (Dodge et al.
1986), and/or hostile-attribution biases and problem solving (Dodge et al.
2006,
2013) facilitate the development and maintenance of antisocial-behavior problems. A broader understanding that social-cognitive and emotional skill development from childhood through adolescence are important for long-term success (e.g., Organisation for Economic Co-operation and Development; OECD
2015).
Several meta-analyses have demonstrated the positive effects of social-skills based programs on reducing aggressive and disruptive behavior (Sandler et al.
2014). For example, two meta-analyses examined the effectiveness of similar interventions to EiE-L which focused on social skills (Beelmann and Lösel
2006). Both studies identified small, but significant effects on anti-social behavior at the post-intervention assessment as well as long-term follow-up (Beelmann and Lösel
2006). Beelman and Lösel (
2006) also examined the effects of these programs in different age groups and found that the effects were largest in adolescence (from age 13 and up; d = .61). In another meta-analysis, Derzon et al. (
2006) detected small but significant effects of social-skills based programs on reducing physical violence and criminal behavior. Similarly, in a meta-analysis by Mytton et al. (
2006), the authors found small but significant post-intervention and follow-up effects in reducing aggressive behavior in high risk youth who attended programs which much like EiE-L focused on developing youths’ social and relationship skills as well as adaptive responses to provocative situations. More recently, in their meta-analysis of prevention studies aimed at enhancing social-emotional skills, Durlak et al. (
2011) identified a small but significant overall effect on problem behavior immediately following treatment, which was maintained at the six-month follow-up. While the primary focus of these meta-analyses was to assess the effects of these programs on externalizing behaviors in adolescence, they also identified positive effects of social-skills based programs on school exclusion, as well as positive outcomes, such as social and communication skills, interpersonal relationships, and school performance. Together these findings suggest that an intervention focusing on building interpersonal communication skills as well as more general social skills may be an efficacious approach to reducing problem behavior and related outcomes, such as school exclusion.
Discussion
How to address student misbehavior is a problem as old as schooling itself. One current strategy in the UK is the use of exclusion from school. Researchers have been concerned with and have called for attention to school exclusion (Gazeley et al.
2015; McCluskey et al.
2015). Owing to this research a lot is understood today about the multiple risk factors for and negative short, intermediate and long term consequences of school exclusions. Programs and interventions have been used to attempt to address these problems, however, to our knowledge rigorous evaluation of the effectiveness of these approaches has been lacking. Evaluation of programs is essential, particularly in the current climate of austerity and reduced government spending, resources should only be directed to programs which are empirically validated and have demonstrated effectiveness. Furthermore, as programs may not only be ineffective but also have the potential to produce harmful effects (Zane et al.
2016) delivering them without rigorous evaluation is risky and unethical. Our research—the first of its kind in the UK—sought to address some of these knowledge gaps by evaluating an intervention aimed at reducing exclusion and problem behavior through improving social communication and broader social skills of students at high risk for school exclusion.
Our results suggested a small but statistically significant difference for self-reported fixed-period exclusion following the intervention. Specifically, at post-intervention, students in treatment schools were
more likely to self-report that they had been excluded in the previous month than students in comparison schools. These results suggest a potential negative effect on school exclusion. In line with common practice across the field of psychosocial interventions and the ongoing debate about the need/utility of adjustments in trials (e.g., Schulz and Grimes
2005), we did not account for multiple outcomes in reporting our findings. However, as our analyses yielded only one statistically significant result, we would interpret this with caution given that the family-wise error rate will be increased. Nonetheless, although not statistically significant, the teacher reported exclusion data as well as the official records of exclusion revealed the same direction of effects, more exclusions in the treatment schools post-intervention. With respect to the secondary outcomes, our analyses revealed no significant differences between the students in the treatment schools versus control schools on any of the 15 outcomes, including communication skills, prosocial behaviors, student–teacher relationships, antisocial behavior, delinquency, and official arrest records.
Iatrogenic and null effects are not uncommon in prevention research and researchers have pointed out that understanding of what causes harm in an intervention is as important as understanding what works in order to improve intervention theory and practice (e.g., McCord,
2003). In a recent review of systematic reviews on harmful effects of crime prevention programs, Welsh and Rocque (
2014) identified three key reasons for harmful effects: (1) theory failure; (2) implementation failure; (3) and deviancy training. These are also considered useful for understanding null effects, therefore we considered each reason in turn below.
The theory of change identified by Catch22 for the EiE-L program rested on the link between impaired communication/social skills and behavior, which may lead to exclusions. Although previous studies have suggested links between social-emotional deficits and behavior problems (e.g., Durlak et al.
2011), our findings suggest that targeting social communication and broader social skills may not be an effective strategy to intervene with students at the
highest risk for school exclusion. Even with a short post-intervention follow-up period where we would expect to find the strongest positive relationships, positive effects on communication and prosocial skills were not observed. Further, social communication and broader social skill deficits may simply be symptomatic of other issues rather than causally related to behavior. It is possible, for example, that executive functioning deficits may be confounding this link (see e.g., Hughes et al.
2009). If so, efforts to improve broader social skills would be ineffective if these cognitive deficits were not being addressed at the same time.
The importance of implementation quality and its impact on the success or failure of interventions has been widely demonstrated (e.g., Durlak and DuPre
2008; Wilson et al.
2003). However, measuring implementation quality is difficult because it is a multifaceted construct, which includes the quality of program delivery as well as participant involvement (Bishop et al.
2014). Measures of program delivery include: evaluation of adherence to a curriculum; training of staff; and time spent on/off task in sessions. Participant involvement can include: consideration of attendance or dosage, participants’ engagement, and behavior in sessions. Therefore, with respect to implementation our data shows two areas for concern. First, there is evidence of low exposure to treatment for those in treatment schools. Specifically, from an intended twelve individual and twelve group sessions, the average number of sessions attended was 6.85 for one-to-one sessions and 6.83 for group sessions. This suggests that there were fewer opportunities for the intervention to actually take place than was intended. Given the high-risk sample, low attendance is an understandable challenge, but one that intervention providers should anticipate and for which they should prepare. Problems with attendance and engagement are perhaps more likely when dealing with a high-risk sample.
Second, although Catch22 believe that content was delivered as intended, in other words with high fidelity, and no significant variations were reported by the intervention team, a review of weekly EiE-L session progress and action logs revealed that core workers encountered a variety of organisational and logistical difficulties in several schools. Furthermore, the intervention design allowed for home visits and telephone calls to the students’ families, which could have been employed to address attendance and engagement problems. However, comparatively few phone calls were made (n = 164) and only eleven home visits were completed. For illustration purposes, 47 students did not attend any group sessions at all. If one phone call had been made for each session that these 47 students alone did not attend, then a total of 564 phone calls would have been made. This seems like a missed opportunity for re-engaging youths and their families in the program and in their education more generally. The intervention provider appears to have had low expectations for the attendance and engagement of students, despite aiming to alter their behavior. Poor attendance (dosage) as well as engagement and other relevant factors may affect the impact of interventions (Rothwell
2005).
The third often cited reason for harmful or null treatment effects, deviancy training, has been observed in interventions targeting students with severe behavior problems (e.g., Dishion and Tipsord
2011). The process is often referred to as “deviant peer contagion” (Dodge et al.
2006), “delinquent spiral” (Cécile and Born
2009), or “drift into deviance” (Dishion et al.
1999). Several mechanisms underlying the negative effects of treating students and their behavior problems in a group format have been described. The predominant view is that students in these situations encourage each other’s behaviors through mutual participation and deviant or antisocial talk or verbal statements which are seen as potent sources of reinforcement (Dishion and Tipsord
2011). Developmental psychologists have suggested that children and students who have experienced social exclusion and rejection are more likely to be susceptible to negative group influences in search of belonging; conforming to what the group does and how the majority or a strong individual behaves; and in order to achieve social status (e.g., Gifford-Smith et al.
2005). This is particularly true during adolescence, when students are generally more susceptible to peer influences (e.g., Menting et al.
2015). It is therefore possible that the negative group influences cancelled out the possible positive intervention effect and hence yielded null post-intervention findings.
According to Moon et al. (
2010), “null results, or no differences between groups, are an important but often hidden aspect of scientific inquiry, potentially contributing as much to knowledge as superficially more ‘successful’ studies that support hypotheses and provide positive advances to understanding” (p. 482). There are two possible methodological factors that may account for no effects and so must be considered: measurement problems and statistical power. In terms of measurement, sub-scales from well-validated measures were used and these scales had high reliability in the study sample. In terms of statistical power, the study operated under practical constraints that limited the number of schools/participants. The study was planned on the basis of being able to detect standardized differences of around d = .35 (see Obsuth et al.
2014). The models achieved statistical power very close to that planned and even on occasion bettering it (owing to smaller ICCs than anticipated). Moreover, with the exception of one (student–teacher relationships, from adolescent report data) of the total of 18 tested models, all of the estimates were pointing in the direction of iatrogenic rather than positive intervention effects. This leaves two additional possible reasons for no effects. First, that the intervention was not implemented well enough to result in any change on these outcomes, or second, that the intervention was implemented well, but did not affect the students’ behavior in a meaningful enough way. The relatively high scores on our two measures of implementation quality, students’ behavior in sessions and time spent on-task, suggest that an adequate implementation quality was achieved. However, in the context of relatively low attendance, another often utilized measure of implementation quality (e.g., Durlak and DuPre
2008), it is possible that the treatment providers did not achieve a desired engagement with the program which may have allowed participants to benefit from it. These possibilities are further explored in sub-group analyses presented in Obsuth et al. (in press).
This study suggests that short-term school-based interventions that have not been well-integrated into school provision, or are otherwise ‘external’ to the school, are unlikely to be successful in changing students’ behavior, particularly students who have already had difficulties at school. Whils not ‘news’ to researchers in this field, the intervention approach set out here is one frequently encountered in the real world, particularly when working with students who are marginalised (e.g., Cooper et al.
2007). Implementation of behavioral interventions with high-risk adolescents needs to be carefully managed and teachers need to be on-board from very early on (Nation et al.
2003; Theimann
2016). Adolescence is a developmental period characterised by marked and rapid biological, cognitive, emotional and social changes. As a result, it has been identified as the second major ‘window’ of opportunity for positive changes as well as sensitive period for risk, next in significance to early childhood development (e.g., Moretti and Peled
2004). Given the structural and functional changes in their brain’s dopaminergic system responsible for the regulation of socio-emotional processes, students are more likely to engage in risk-taking behaviors, or behaviors with potential for harm to self and others, such as delinquency, substance use, dangerous driving, than younger children or adults (e.g., Steinberg
2015). They are generally more susceptible to peer influences and are more likely to engage in risk-taking behaviors and/or delinquency in the presence of peers (e.g., Menting et al.
2015). Interpersonally, students expand their social circles; spend more time with peers and form their first serious romantic relationships. In their apparent striving to establish a new balance between dependence on their carers for support and their autonomy or independence (e.g., Oudekerk et al.
2015), it may appear that they no longer rely on their parents and other significant adults (such as teachers, mentors) for help and support. However, evidence suggests otherwise. Recent studies highlight the importance of positive student–teacher relationships and strong school bonds in healthy adolescent development (Silva et al.
2016; Theimann
2016). For example, Theimann (
2016) found that positive student–teacher relationships in the context of positive bonds to school were related to lower rates of delinquency in students from age 13 to 16. A meta-analysis by Wilson et al. (
2003) found that interventions delivered by teachers were more effective than those delivered by offsite providers. Anecdotal evidence from the EiE-L core workers indicated that in some instances schools informed students that they were enrolled on the intervention because they were the “worst kids”; this may not only hinder any engagement in intervention but also jeopardise the teachers’ relationships with the students and thus contributed to negative effects. Adolescence is a volatile transitional period and more care should be taken to consider this when introducing and delivering any intervention. Moreover, positive experiences and relationships within schools (both with peers and teachers) have been well documented (e.g., Layard et al.
2014; Silva et al.
2016; Theimann
2016), therefore the tendencies to exclude are particularly troubling.
Rates of exclusion were alarmingly high for the students in this study, with 30–50 % (based on official records and questionnaires, respectively) receiving a temporary exclusion in both treatment and control schools in the year prior to the study. Moreover, nine per cent of students in treatment schools and 6 % of students in control schools experienced an officially recorded exclusion in the six week period immediately following the intervention. These rates were much higher based on teacher and adolescent reported exclusions. This discrepancy may reflect the often described problem of unrecorded/unreported school exclusions (e.g., Gazeley et al.
2015). Furthermore, multiple exclusions were not uncommon within the students who were included in our analyses, suggesting that the study had indeed correctly sampled those at the greatest risk of exclusion.
The rates at which exclusions occurred among our sample suggest that schools are struggling to deal with a significant proportion of students for whom they are responsible. The need to think differently about how to manage students with problem behavior is clear. An approach that emulates the collaborative emphasis of the Communities that Care (Kim et al.
2015) or Positive Behavioral Interventions and Supports (e.g., Horner and Sugai
2015) models, with several care providers sharing the responsibility for tackling a problem, may prove fruitful in this respect. These approaches suggest that, to achieve effective and lasting change in students’ interactions, behaviors, and emotions, the whole school needs to be addressed, with the empirically supported view that those with greatest need will benefit most (Tolan et al.
2004). It should be noted that such approaches incorporate greater assistance and resources for those with the greatest needs, but within an inclusive, whole-school framework. Programs employing similar principles of care have been evaluated and revealed positive effects on youth behavior, delinquency as well as school exclusions in the UK and elsewhere (e.g., Pritchard and Williams
2001). Moreover, in further support of the importance of a school-wide supportive context, recent longitudinal research (Layard et al.
2014) revealed that good experience of education was more important than what individuals learned at school for good outcomes in life 40 years later. This study highlights a need for a shift from a focus on achievement alone to a focus on healthy child development in schools. Efforts should be made to identify feasible alternatives building on principles of inclusion and healthy emotional development. Policy change may be required to allow schools sufficient autonomy to deliver such models.
As with any project, this study has some limitations which were mainly related to scheduling. Information was collected from participants at particularly busy times at the beginning and end of the school year. This contributed to initial difficulties completing all of the baseline data collection in September and led to the two phase design. As we noted above, the providers self-reported program fidelity, in other words that the sessions were delivered as intended. Provider reports of problems in the implementation of an intervention are a common practice in assessing treatment fidelity in behavioral program evaluation research (see for example a recent overview of meta-analyses by Sandler et al.
2014). We have built on the recommendation of Sandler et al. (
2014) by broadening the assessment of implementation quality, adding measures of dosage and treatment engagement. However, in this study the relatively low attendance and utilisation of family contact, suggests a gap between provider perceptions of implementation and achieved program delivery. Independent observations of the implementation quality would perhaps lend a better, less biased, insight into the implementation process. However, due to the short time span of the current evaluation (2 years in total) this was not possible in this study. Similarly, our resources and time restrictions did not allow for the collection of information related to therapeutic alliance, which could further help us explain our findings.
Yet, this study also has several important strengths. It is one of the first school-based large-scale independent c-RCTs of a behavioral intervention aimed at high-risk students in the UK. The research involved an innovative approach to fieldwork in recruiting a large temporary fieldwork team in order to collect data more efficiently than would be possible with a smaller but full-time group of fieldworkers. Given the high-risk group, those typically absent from school surveys, a relatively high retention rate was achieved (77 % of n = 606 at baseline), which is notable given the nature of this population: it is well-known that individuals with the highest levels of problematic behavior are at particular risk of non-participation and dropping out (e.g., Eisner and Ribeaud
2007). Questionnaires were administered to multiple informants and official records were secured (for the primary outcome), which allowed the interpretation of findings with greater confidence. An effective collaboration between the agencies involved in the study; namely the Greater London Authority, the Education Endowment Foundation, Catch22, I CAN and the 36 schools; and the research team was developed. We achieved good balance on the key variables identified in the protocol. We also included additional measures that allow us to understand this high-risk group of students and the project allows the possibility of tracking this group over the long-term using administrative data. Finally, we achieved sufficient power to detect small to medium effects.
While this study focused on schools in England, we believe the findings are generalizable to other jurisdictions. The schools included in this study represented deprived areas of London where the social composition does not necessarily reflect that of the rest of the UK or other countries. However, trends in exclusion appear to be relatively universal across countries, with boys, ethnic minority students and those with stated, as well as undiagnosed, learning difficulties being disproportionately suspended and excluded from schools (Achilles et al.
2007; Day-Vines and Terriquez
2008; Losen et al.
2015). Studies focusing on whole-school approaches to tackle school exclusion have had equal success in the US (Bradshaw et al.
2010) and other countries, such as Norway (Sørlie and Ogden
2015). There are significant differences in approach to the provision of public and social services in these two countries, as well as in the historical and cultural context of the integration of ethnic minorities and social classes. However, despite these differences, similar approaches to tackling shared problems appear efficacious. This suggests that research in this area will be applicable to different cultural contexts.