Special needs education crystallises in various shapes in different parts of the world, going from fully inclusive education through fully separate and specialized institutions. It is aimed at giving appropriate education to pupils with special educational needs, but remains a controversial practice everywhere (Carlberg & Kavale, 1980).

First of all, special education is costly to provide (Greene & Forster, 2002). For instance, in Flanders, the Dutch part of Belgium where the present study was conducted, the educational expenditure per pupil in special education is three times as big as the expenditure per pupil in mainstream education (Scheys, 2011). Intelligibly, most governments want to know whether the benefits in terms of student outcomes outweigh the extra costs, especially at a time where more and more pupils are diagnosed with a label making them a candidate for special education (Graham & Jahnukainen, 2011).

Second, previous research has provided little convincing evidence that supports or rejects the effectiveness of special or mainstream school placement for pupils with special needs (Ainscow et al., 2006). Contradictory findings have been reported for different outcomes such as academic performance, social adjustment, and personality traits, as well as for different target groups such as pupils with emotional or behavioural problems, mild intellectual disability, or learning difficulties. Moreover, the existing literature is out-of-date. The complexity of special education as a field as well as the fact that most studies so far show methodological shortcomings (e.g., not accounting for selection bias and dependence among observations) might partly explain these inconsistencies (Odom et al., 2005).

In Flanders, separate special education schools cater for pupils with disabilities who cannot be accommodated within mainstream education. Besides, as in most countries, more and more proponents of integrated or inclusive education are found since the nineties, among policymakers as well as among researchers and practitioners. As a result, a part of the special-needs pupils today go to mainstream schools but receive extra support from special education (EACEA, 2010).

Randomly assigning pupils to special education (i.e., treatment group) versus mainstream education (i.e., control group) would not be very ethical. As a consequence, most previous studies have relied on simple contrasts between pupils who were referred to special education and those who were not. However, in such case, selection bias is particularly likely to occur (Graham & Kurlaender, 2011). We chose a quasi-experimental design (propensity score-matching) to meet these methodological shortcomings (Morgan, Frisco, Farkas, & Hibel, 2010).

We narrowed our focus to pupils with social, emotional, behavioural, or learning difficulties and thus excluded pupils with hearing, vision, physical, and severe intellectual impairment from our sample, since it is very difficult to find a match for these pupils.

Research question:

‘Do pupils who are referred to special education acquire greater or smaller math skills during the first two years of primary education than pupils with similar characteristics who go to mainstream education?’ In other words: ‘Is referral to special education at the age of six (after kindergarten) leading to higher or lower math scores in the following years?’

Ainscow, M., Booth, T., Dyson, A., Farrell, P., Frankham, J., Gallannaugh, F. et al. (2006). Improving Schools, Developing Inclusion. London: Routledge. Carlberg, C., & Kavale, K. (1980). The Efficacy of Special Versus Regular Class Placement for Exceptional Children: a Meta-Analysis. The Journal of Special Education, 14(3), 295-309. Dehejia, R. H., & Wahba, S. (2002). Propensity Score-Matching Methods for Nonexperimental Causal Studies. Review of Economics and Statistics, 84(1), 151-161. doi: 10.1162/003465302317331982 EACEA. (2010). Organisation of the education system in the Flemish Community of Belgium The Education, Audiovisual and Culture Executive Agency, European Commission. Retrieved from http://eacea.ec.europa.eu/education/eurydice/documents/eurybase/eurybase_full_reports/BN_EN.pdf Graham, L. J., & Jahnukainen, M. (2011). Wherefore art thou, inclusion? Analysing the development of inclusive education in New South Wales, Alberta and Finland. Journal of Education Policy, 26(2), 263-288. doi: 10.1080/02680939.2010.493230 Graham, S. E., & Kurlaender, M. (2011). Using propensity scores in educational research: General principles and practical implications. The Journal of Educational Research, 104, 340-353. doi: 10.1080/00220671.2010.486082 Greene, J. P., & Forster, G. (2002). Effects of Funding Incentives on Special Education Enrollment New York, NY: Manhattan Institute for Policy Research. Retrieved from http://www.manhattan-institute.org/pdf/cr_32.pdf Morgan, P. L., Frisco, M. L., Farkas, G., & Hibel, J. (2010). A Propensity Score Matching Analysis of the Effects of Special Education Services. The Journal of Special Education, 43(4), 236-254. Odom, S. L., Brantlinger, E., Gersten, R., Horner, R. H., Thompson, B., & Harris, K. R. (2005). Research in special education: Scientific methods and evidence-based practices. Exceptional Children, 71(2), 137-148. Scheys, M. (2011). Flemish education in images 2010-2011. Retrieved from Flemish authorities http://www.ond.vlaanderen.be/publicaties/eDocs/pdf/91.pdf