Education that fits: Review of international trends in the education of students with special educational needs

Publication Details

The purpose of this review is to outline international trends in the education of students with special educational needs, with the aim of informing the Ministry of Education’s current review of special education.

Author(s): David Mitchell PhD, College of Education, University of Canterbury, for the Ministry of Education.

Date Published: July 2010

Please consider the environment before printing the contents of this report.

This report is available as a download (please refer to the 'Downloads' inset box).  To view the individual chapters please refer to the 'Sections' inset box.  For links to related publications/ information that may be of interest please refer to the 'Where to Find Out More' inset box.

Chapter 12: Non-inclusive educational settings

Obviously, the reciprocal of inclusive education, which was addressed in the previous chapter, is non-inclusive education. Therefore, many of the issues that were traversed in that chapter have relevance for the present one. In particular, the evidence that related to student outcomes in inclusive education was usually compared with outcomes in some form of non-inclusive settings, such as special schools or units.

In this chapter, the focus will be on the following non-inclusive educational approaches: special schools, special classes/units, streaming, setting, within-class ability grouping, and individual instruction, some of which are used in combination.

12.1 The ‘Where to Learn’ Dilemma

As Shaddock et al. (2009) pointed out, debates about what constitutes an appropriate setting for SWSEN have had a long and turbulent history, dating back at least to the seminal article of Dunn (1968). These debates illustrate what Norwich (2008) referred to as the ‘where to learn’ dilemma. As indicated in Chapter Eleven, the value of various placements, from segregation to total inclusion, has been interrogated on ideological, philosophical and empirical grounds. For example, strong supporters of special education (and, by inference, non-inclusive settings), Kauffman and Hallahan (2005) made the following case:

Since its inception, special education has been conceptualised as special instruction. But those who invented special education recognized that special instruction sometimes requires a special place, simply because no teacher is capable of offering all kinds of instruction in the same place and at the same time and that some students need to be taught things that others don’t need. So, as has been recognised all along, the specialized places in which special education sometimes occurs are necessary for special instruction, especially if it is to be done well. There is no magic in any place, either the regular classroom or a special class. Place, by itself, does not represent good special education. Special education is neither good nor bad because of where it is offered. The instruction is what matters and what makes special education (p.63).

12.2 Special Schools and Special Classes

Special schools are normally considered to be schools that cater exclusively for SWSEN with severe learning difficulties, physical disabilities, sensory disabilities, behavioural problems or multiple disabilities. Students attending such schools generally do not attend any classes in mainstream schools. They are usually specifically designed, staffed and resourced to provide the appropriate special education and related services for SWSEN. Qualifiers to all of the foregoing were used deliberately for, as we shall see, the character of special schools is undergoing considerable changes in many parts of the world. Special classes/units (sometimes referred to as ‘self-contained classrooms’ in the US) are normally considered to be separate rooms dedicated solely to the education of SWSEN within a larger school. Such classrooms are typically staffed by specially trained teachers who provide individualised or group instruction to students with a particular disability.

Firstly, let us consider some of the statistics on special school and special class/unit placements. The OECD (2005) has presented a comprehensive set of data on educational provisions for SWSEN in 31 countries for around 1999-2003. These are shown in Table 12.1. Several points should be mentioned:

  1. The data related to different age groups, as the compulsory starting age for school differs across countries.
  2. ‘Segregated provisions’ referred to special schools and fulltime, or almost fulltime, special classes.
  3. The varying percentages of SWSEN (from a low 0.9% in Greece to a high of 15.0% in Iceland) reflected different definitions of such students. For example, in England the 3.2% of SWSEN referred only to students with statements; another 13.8% were identified less formally as having special educational needs, while Sweden did not gather data for SWSEN who were fully included.
  4. The percentages of SWSEN in non-inclusive settings ranged from several countries with less than 1% (Cyprus (0.7%), Greece (0.5%), Iceland (0.9%), Italy (under 0.5%), Norway (0.5%), Portugal (0.5%) and Spain (0.4%)) to several with 4-6% (French –speaking Belgium (4.0%), Dutch-speaking Belgium (4.9%), Czech republic (5.0%) Germany (4.6%), and Switzerland (6.0%)).
  5. The likely fluidity of these provisions must be noted. For example, non-inclusive placements in the Netherlands had fallen sharply compared with a few years before the period portrayed in Table 12.1 as a result of changes in legislation (see Chapter Seven, section 7.4.1). Also, there is some evidence that the Swedish figure might have under-represented the later situation of a rising number of SWSEN attending special schools (Emanuelsson et al., 2005).

Table 12.1: Provisions for SWSEN (OECD data)
CountryNo of Compulsory School-aged pupils
% of SWSEN
% of students in segregated provision
Year of Reference
Austria848,1263.2% 1.6%2000/2001
Belgium (DE)9,4272.7%   2.3%2000/2001
Belgium (F)680,360 4.0%4.0%2000/2001
Belgium (NL)822,666 5.0%4.9%2000/2001
Czech Repub1,146,6079.8%5.0%2000/2001
Denmark670,000 11.9%1.5%2000/2001
England9,994,159 3.2%1.1%1999/2000
France9,709,000 3.1%2.6%1999/2000/2001
Germany9,159,0685.3%4.6% 2000/2001
Greece1,439,411 0.9% < 0.5%1999/2000
Hungary1,191,7504.1%3.7% 1999/2000
Italy8,867,8241.5%< 0.5%2001
Norway601,8265.6%0.5% 2001
Poland4,410,516 3.5%2.0%2000/2001
Portugal1,365,8305.8%< 0.5%2000/2001
Slovenia189,342 4.7%(:)2000

As noted by Riddell et al. (2006), countries differed in their placement of SWSEN, according to the three-way classification described in Chapter Three, section 3.1 of the present review. Overall, for reporting countries in another set of OECD data, they observed the following:

  • Category A (disabilities): there was considerable variation across countries, between a preference for regular classes (Canada (New Brunswick)) to a preference for special schools (Belgium (Flemish Community)). Most countries had a mix of the three types of placements (e.g., US, Turkey, France, Slovak Republic, Japan, Hungary, Czech Republic and Korea).
  • Category B (difficulties): there was a considerable variation across countries, between a preference for regular classes (Canada (New Brunswick) to a preference for special schools (Belgium (French Community).
  • Category C (disadvantages): there was a definite preference for regular classes in all countries.

When one drills down into country statistics, further interesting patterns emerge. For example, in England, there is clear evidence that not only are fewer students being educated in special schools (1.1% in 2003, compared with 1.5% in 1983, according to the Pupil Level Annual Schools Census in 2003), but the population of special schools is undergoing change. Recent data from that country shows a gradual increase in the number and percentages of SWSEN attending special schools as having behavioural, emotional and social difficulties (BESD) and autistic spectrum disorder (ASD), as can be seen in Table 12.2. This table shows that the two categories combined constituted 30.9% of the special school population in 2008, compared with 25.0% in 2005, with the greatest increase being recorded for students with ASD (from 11.1% to 16.0%).

These data reflect the rapid increase in the number of young people receiving a diagnosis of ASD and BESD in all jurisdictions of the UK as documented by Lloyd (2003) and Pirrie et al. (2006).

Table 12.2: Special schools and number and percentage of SWSEN by type of need
YearType of Need
N=%N= %Total %
200512,47013.9 9,90011.125.0



Another example of an analysis of the population of SWSEN in non-inclusive settings is embedded in US data for 1995, which shows the distribution of students by the number of their disabilities. This information is outlined in Table 12.3.

Table 12.3: Number and percentage of students receiving special education and related services in various educational environments, by number of disabilities in the US
Number of DisabilitiesRegular classroom settingResource RoomSeparate ClassMore than one of these locationsTotal
Percent   28.0   36.3   20.6   15.1   100.0
Two or more


  1. Source: 22nd Annual Report to Congress (U.S. Department of Education, 2000), which acknowledges the 1995 National Health Interview Survey.
  2. Special day schools, special residential schools, homes, hospitals or institutions, were excluded due to small sample sizes.      

The Department of Education drew attention to the following (a) a larger percentage of children with co-occurring disabilities than of students with one disability received their educational services in a separate classroom located in a regular school (for either all or part of the day) (b) students with only one disability received their educational services primarily in a resource room located in a regular school, and (c) compared to students with only one disability, a greater proportion of those with two or more co-occurring disabilities received services in more than one of the specified locations (31% compared to 15%).

Another interesting pattern emerged in an analysis of the influence of population density on the percentage of students being educated in non-inclusive settings, carried out in Europe by Meijer et al. (2003). They found a high correlation between these two variables (0,60, N=15 countries). In other words, ‘about 36% of the variance of the percentage of segregated children is explained by population density’ (p.80). The authors explained this finding in terms of the disadvantages of special school placements in countries with low population density: greater travel distances, negative social consequences as children are taken out of their social environments, and the higher costs incurred.

12.3 New Roles for Special Schools

In their recent review of special education in the ACT, Shaddock et al. (2009) noted that special schools accounted for 0.9% of students in public schools or 0.5% of the total government and non-government enrolments. They went on to propose new roles for some special schools, and different models for meeting the needs of students who currently attend them. Their rationale was to ‘(a) capitalise on the expertise and resources in these facilities; (b) extend the schools’ connections with their communities and surrounding schools; (c) reduce travel for students with disabilities; and (d) give students the opportunity to receive an appropriate education (including school friendship opportunities) in their own neighbourhood’ (p.17).

Shaddock et al. presented quite a lengthy review of possible new roles for special schools, making the following points:

  • In the UK, Warnock (2005) encouraged special schools to become ‘specialist schools’, offering services to a broader section of the school population.
  • The NSW Public Education Inquiry (2002) encouraged special schools to form linkages with regular schools, suggesting that teachers in special schools could accept roles as co-ordinators to assist regular schools with inclusion, sharing resources and their expertise with teachers and assistants and providing outreach services.
  • Innovative practices documented by Farrell (2008) and by Gibb (2007) included suggestions that ‘exemplary special schools’ could share best practice in:
    • teaching multi-age and diverse classes,
    • mentoring and working collaboratively with regular schools,
    • training teachers and assistants how to differentiate work,
    • teaching specific skills to students individually and in groups,
    • developing individual learning and behavioural programmes,
    • providing outreach services to support the integration, transition or the enrolment of students with disabilities through information on the student or the impact of the disability on the student’s capacity to learn,
    • developing individual programmes for students,
    • assessing students for assistive technology,
    • screening the speech and language of students,
    • establishing new special units in regular schools,
    • organising parent information sessions, IEP meetings and visits from professionals to support their mainstream colleagues,
    • offering specialist college-level vocational courses on car repairs, hospitality, building, sport and gardening to students and adults after school hours
    • offering short-term placements to students to develop an effective behaviour management programme, with ongoing support when the student returns to the regular school.

In a similar vein, an earlier report from the European Agency for Development in Special Needs Education (2003) noted a trend in European countries in which special schools and institutes were being transformed into resource centres, with such functions as (a) training teachers and other professionals, (b) developing and disseminating materials and methods (c) supporting mainstream schools and parents, (d) providing short term or part-time help for individual students, and (e) supporting students to enter the labour market. (See Chapter Fourteen, section 14.1 for more details on eight European countries in which teachers from special schools are utilised to support regular class teachers.)

In England, the 1997 Green Paper, Excellence for All Children, signalled the government’s commitment to inclusive education and the need to rethink the role of special schools within that context. The subsequent document, Every Child Matters (Department for Education and Skills, 2003) envisaged special schools as ‘providing education for children with the most severe/and complex needs and sharing their specialist skills and knowledge to support inclusion in mainstream schools’ (p.26). They would pursue the latter role through regional centres of expertise to be developed in association with local authority support services. This could be achieved by setting up ‘federation, cluster and twinning arrangements with their mainstream counterparts’ (p.35). According to Farrell (2008), a schools’ building programme scheduled for 2016-2021 will enable secondary schools to have specialist facilities and schools contained within or adjacent to them, which will facilitate relationships between special and regular schools. Already, educational authorities have established a specialist schools programme involving more than 50 special schools. Each school specialises in one area: cognition and learning; communication and interaction; physical and sensory; or behavioural, emotional and social difficulties and has been allocated the necessary time, funding and resources to share their expertise and resources with other schools, agencies, services and the community (Farrell, 2008). However, with the recent change of government in the UK, it will be interesting to see how special schools fare in the future. Some indication of what might occur can be found in a recent Conservative Party commissioned report (Balchin, 2007), which included the comment that ‘The saddest and most serious result of the present Government’s Inclusion policy has been the closure in the last decade of special schools and the concomitant destruction of special school places’ (chapter Six). The report went on to ‘demand not just a moratorium on the closure of special schools, but also an active exploration of how we might recreate the number of places that have been destroyed’ (ibid.).

In Sweden, too, special schools are being transformed. In 2001, all special schools, except for those providing sign language education, were re-designated as special needs resource centres. These were being developed to support inclusion in mainstream classes. A specialist teacher working as a member of the mainstream school staff mainly provides support. Municipalities are responsible for ensuring that necessary expertise is available and may request support from the Swedish Institute for Special Needs Education (European Association for Development in Special Needs Education, 2003).

When it emerged from the apartheid era, South Africa was determined to create special needs education as a non-racial and integrated component of its education system. In a 2001 White Paper (Department of Education, 2001, 1:14), several findings of commissions on special needs education were reported. These included: policies aimed, inter alia, at bringing about qualitative improvements in special schools and their phased conversion to resource centres and the establishment of district-based support teams (Department of Education, 2001).

12.4 Research into Non-Inclusive Settings

As mentioned in the introduction to this chapter, the evidence that related to student outcomes in inclusive education was usually compared with outcomes in some form of non-inclusive settings, such as special schools or units. This won’t be repeated here, apart from reiterating the conclusion that ‘the evidence for inclusive education is mixed but generally positive, the majority of studies reporting either positive effects or no differences for inclusion, compared with more segregated provisions’.  

Even so, some writers continue to argue for special units and classes for students with particular disabilities, for example, students with learning disabilities, those with ASD and students with profound sensory impairment. Some research support can be found for this case in Swanson & Hoskyn’s (1998) report on 180 interventions with students with learning disabilities that found a slight benefit for some students in ‘pull-out programs’. However, the researchers explained the benefits in terms of the quality of the instruction rather than where it was provided.

Shaddock et al. (2009) have summarised other arguments in favour of non-inclusive settings. Thus, they drew attention to writers who argue that regular classrooms may not be set up to assist students with ASD, many of whom need specialised curricula and teaching approaches (Mesibov & Shea, 1996; Sainsbury 2000). They also noted that despite the lack of evidence for the beneficial effects of non-inclusive placements on learning, many Australian parents continue to want more special units in primary and secondary schools, not fewer (Nitschke & McColl, 2001) and that reviews have shown that parents and teachers strongly support a continuum of services (McRae, 1996; NSW Public Education Inquiry, 2002; Commonwealth of Australia, 2002; Nitschke & McColl, 2001). Again according to Shadock et al., parents want the option to move their child to a special education setting if the regular class proves to be problematic, and the inclusion of some students has certainly proved to be problematic for some sectors (Department of Education and Training Western Australia, 2001). Parents and teachers have reported bullying, peer rejection, inappropriate curricula, failure/inability to differentiate, lack of teacher time, inadequate teacher training, limited funding and resources, students with disabilities being taught by assistants - especially in secondary schools (Commonwealth of Australia, 2002). Conversely, some parents speak in appreciation of special schools, citing such advantages as positive expectations, ease of administering medicines, fully accessible physical environments, better behaviour management, and access to specialists (Department for Education and Skills, 2003). As Warnock (2005) pointed out, too, students with disabilities may be excluded socially and academically in a regular school and so special schools may be the salvation for many students. Indeed she went so far as to describe inclusion as ‘possibly the most disastrous legacy of the 1978 Report’ (p.20), claiming that ‘There is increasing evidence that the ideal of inclusion, if this means that all but those with the most severe disabilities will be in mainstream schools, is not working’ (p.32). And, finally, Shaddock et al. noted that another rationale for the continued existence of special schools or classes may be, as suggested by Sorrells et al. (2004), that separate classes for ‘difficult to teach’ children may function as a safety valve for schools rather than as a preferred place of learning for students. These authors further suggested that specialised programmes may simply be part of the repertoire that public schools have for dealing with problems.

Thus, one has to look beyond the empirical evidence of educational efficacy to other more complex motivations for justifying the retention of non-inclusive educational settings.

12.5 Ability Grouping

There are two aspects to placing learners in ability groups: (a) ability grouping between classes, sometimes referred to as ‘tracking’ or ‘streaming’ or ‘setting’, and (b) ability grouping within classes. The relevance of such grouping for SWSEN is that they are highly likely to be placed in ‘lower streams or groups, thus constituting  a form of de facto segregation.

According to Benn & Chitty (1996), most secondary schools in the UK employ some form of ability grouping, usually setting, for at least some subjects, while in the US, tracking in various forms has been among the predominant organising practices in public schools for the last century (Rubin, 2008).

In a recent review, Duckworth et al. (2009) concluded that ’much of the available evidence suggests that the effects of ability-grouping on pupil attainment is limited and no firm conclusions can be drawn from its use’ (p.30). This conclusion reflects the results of a meta-analysis carried out by Lipsey & Wilson (1993), who reported on the impact on learners’ achievement of within-class ability grouping and between-class ability grouping. Their results showed a negligible overall effect size of less than 0.10, with a range of -0.03 to 0.22. A similar result was reported by Hattie (2009), whose meta-analysis yielded an effect size of 0.21 for ability grouping’s impact on student achievement. In other words, these two reviews showed that ability grouping had little or no significant impact on student achievement. Unfortunately, in neither of these reviews were separate results reported for SWSEN.

A recent Dutch review of the literature, however, did differentiate between high- and low-achieving learners (Houtveen & Van de Grift, 2001), It concluded that although the mean results of studies showed higher achievement in ability groups than in mixed-ability groups, this was mainly due to the fact that high-achieving students benefitted more than low-achieving students. The authors cited several studies where low-achieving learners performed more poorly in between-class ability groups than in mixed-ability groups (e.g., Gamoran, 1992; Hallam & Touttounji, 1996).

There is evidence, too, that ability-grouping practices may widen gaps in achievement, with students in high-ability streams doing better than in mixed-ability groups, while placements within low-ability groups has a negative impact on student attitudes towards school and their motivation and achievement (Duckworth et al. 2009; Feinstein & Symons, 1999, Robertson & Symons, 2003). Also of relevance is an early UK study by Fogelman et al. (1978), which found that in comprehensive schools with mixed ability grouping practices, a higher proportion of lower attaining students were entered for national examinations.

Ability grouping is not an all or nothing idea, for it is possible to have ability groups for some subjects and mixed ability groups for others. This arrangement is sometimes referred to as ‘setting’.

Another drawback of ability grouping, as indicated by Duckworth et al. (2007), is that although the importance of students being able to move sets (in the UK) has been stressed, in practice there is very little movement, even when teachers become aware that students are wrongly allocated. Another interesting finding reported by Duckworth et al. was that among secondary school students studying mathematics in ability-grouped sets, 83% either wanted to return to mixed ability sets or to change their set. Their own research with over 8,000 students in 45 secondary schools also showed that a high proportion of them were unhappy with their set or class placement. For example, in mathematics, where there was the highest level of ability grouping, 38% were unhappy with their set or class placement; unsurprisingly, more students in the bottom set (62%) wished to change their set. Significantly, their reasons for wanting to change were more related to learning than status. For many of them they felt there was a mismatch between the work set and what they perceived was appropriate.

A UK study investigated the effects of setting in English, mathematics and science on the academic self-concepts of secondary school learners (Ireson et al., 2001). The results showed that students’ self-concepts were higher in schools with moderate levels of setting. It was also found that the degree of setting in mathematics and science had no effect on academic self-concepts, but setting in English tended to lower the self-concepts of the higher attaining learners and raise the self-concepts of lower attaining learners.

In summarising their interpretation of the research, Houtveen & Van de Grift (2001), put forward a range of arguments as to why ability grouping is detrimental to low-achieving learners:

  • Being assigned to low-ability groups communicates low expectations to students, which might be self-fulfilling.
  • Because ability groups often parallel social class and ethnic groupings, they may increase divisions along class and ethnic lines.
  • Between-class ability grouping reduces students’ opportunities to move between groups.
  • Low-achieving students tend to receive less instruction when placed in ability groups than when placed in mixed ability groups.
  • Ability groups composed of low-achieving students do not provide a stimulating learning environment and lack positive role models.

In a similar vein, MacIver et al. (1995) pointed out that in US research there is evidence that ‘low-track’ classes are much more likely to receive course content that focuses on below-grade level knowledge and skills than high-track classes.

In reporting the results of two meta-analyses that examined the impact of ability grouping and mixed ability grouping on student learning at the elementary and secondary school levels, Slavin (1996) drew the following conclusions:

  • use mixed ability groups for most content areas;
  • encourage learners’ identification with mixed ability groups in order to promote acceptance of diversity; and
  • use ability-grouping only when it will increase the efficacy of instruction or provide more time for instruction on a specific skill.

Due to the disadvantages of streaming (or tracking) outlined above, many schools in the US are implementing what is referred to as ‘detracking’, which involves students being deliberately positioned into classes of mixed ability (Rubin, 2008; Argys & Rees, 1996).

Finally, the inconsequential impact of separate settings on the educational outcomes of most students, together with the negative effects on SWSEN, have refocused researchers’ attention on the variables that do make a difference, many of which are summarised in Chapter Ten. Once again, as Shaddock et al. (2009) emphasised, ‘the research refocuses attention on one critical variable – how teachers teach in their own classrooms’ (p.86, emphasis in original).

12.6 Individual Instruction

As noted by Shaddock et al. (2009), a research synthesis by the Best Evidence Encyclopaedia (BEE) of approaches for helping struggling readers found that classroominstructional approaches produced effect sizes of over 0.5, while one-to-one tutoring by teachers, paraprofessionals and by volunteers produced effect sizes of 0.38, 0.24 and 0.16, respectively. Similarly, Hattie (2009) concluded that, ‘The evidence supporting individualised instruction…is not so supportive’ (p.198). These finding seem counter intuitive: surely individual instruction should be better! Obviously, the social context of the classroom is an important contributor to learning and the need for resource-intensive one-to-one instruction should be reconsidered (Shaddock et al., 2009).

12.7 A Final Word

After their thorough review of non-inclusive educational settings, Shaddock et al. 2009) arrived at a conclusion that is supported by the present writer:

Leading practice does not strongly support the further development of separate placements for students with a disability, in general. As the logic supporting separate provision – preparing students to take their place in society by educating them separately - is somewhat elusive, and as separate placements are not strongly supported by empirical research, the case for such placements should always be the one to be argued (p.87).

Ultimately, to quote Shaddock et al. (2009) again:

…the development and continuation of such [separate] programs should be based on the extent to which they improve student learning outcomes in ways valued by the students, parents and carers, and teachers. Data and evidence, not conviction and ideology, are the key considerations (p.16).

The same criteria should, of course, apply to inclusive educational programmes, indeed to all teaching strategies, as argued throughout this review.

12.7 Summary

  1. The evidence related to student outcomes in inclusive education is usually compared with outcomes in some form of non-inclusive settings.
  2. Non-inclusive educational settings range from special schools, through special classes/units and various forms of ability grouping, to individual instruction.
  3. The ‘where to learn debate’ has been interrogated on ideological, philosophical and empirical grounds.
  4. According to OECD data, the percentages of SWSEN in non-inclusive settings range from several countries with less than 1% to several with 4-6%.
  5. There is evidence that the population of special schools is undergoing change. For example, recent data from England shows a gradual increase in the number and percentages of SWSEN attending special schools as having behavioural, emotional and social difficulties and autistic spectrum disorders.
  6. Many countries are developing new roles for special schools by converting them into resource centres with a range of functions replacing direct, full-time teaching of SWSEN.
  7. Despite the lack of evidence for the beneficial effects of non-inclusive placements on learning, many parents and teachers strongly support a continuum of services, including special schools and units.
  8. Research into ability grouping shows that, overall, it has little or no significant impact on student achievement, although high-achieving students appear to benefit more than low-achieving students, who suffer from disadvantages in being placed in low ability groups.
  9. Paradoxically, individual instruction has a low impact on student achievement, suggesting that the social context of the classroom is an important contributor to learning.
  10. A fitting conclusion would be that the continuation of non-inclusive educational settings should be based on the extent to which they improve student learning outcomes in ways valued by the students, parents, and teachers. Data and evidence, not conviction and ideology, should be the key considerations.