Costs and Effects of the Step By Step Roma Special Schools Initiative

07 November 2002

Benjamin Levin¹

Abstract

An analysis was conducted of the financial impact (costs and effects) of the Step by Step Roma Special Schools Initiative. This initiative in Bulgaria, the Czech Republic, Hungary and Slovakia involved measures aimed to support the moving of Romani pupils from schools for the mentally handicapped – "special schools" – to a mainstream curriculum in mainstream schools.

A review of related literature found that:

• Special school settings are generally much more expensive than mainstream settings;
• Integration is generally regarded as a preferable alternative, especially where students do not have severe disabilities; and
• Improved educational outcomes yield very substantial lifetime benefits, both tangible and intangible, to individuals and society.

All of the above supports integrating Romani children into mainstream schools and classes.

A comparison was carried out in four pilot schools – one in each country – of the costs of current special school sites compared with implementation of an integrated system² with strong support for student success in mainstream schools. The integrated approach would include reasonable class sizes (approximately 20 students), strong training support for teachers, extensive use of Romani assistants, access to special education teachers and other supports. All other benefits to schools and to Romani students and families would remain unchanged to ensure that there were no perverse incentives for continued segregation.

The analysis found that a strong integrated programme could be provided with approximately the same resources currently allocated to special schools and classroom programs. This is for the most part because class sizes in special schools are small, averaging around 10, whereas the students could be successful in larger classes with appropriate supports.

There is also good reason to expect significant long-term social and economic benefits from the integration approach.

The author's conclusion is that the existing evidence supports the cost-effectiveness of integration with support into mainstream programmes and schools as a desirable approach to the education of Romani children in the four countries. This analysis supports and strengthens the case for integration based on human rights and human dignity arguments.

Costs and Effects of the Step by Step Roma Special Schools Initiative

The Step by Step Roma Special Schools Initiative, a programme of the Open Society Institute (OSI), worked with schools, Romani children and their families in four countries – Bulgaria, the Czech Republic, Hungary and Slovakia. In these countries, a large proportion of Romani children are diagnosed as mentally disabled and placed in special classes or schools. Their educational outcomes in these programmes are entirely unsatisfactory. There is pressure on countries to change this situation. The purpose of the Initiative was to show that Romani children can be successful in the mainstream curriculum given appropriate support. An analysis of the costs and benefits of integration was seen as important for countries that might be looking at substantial long-term changes in policy regarding the education of Romani children.

Theoretical Approach

Many educational innovations are undertaken with little or no attention to their short- or long-term costs. Even more rare is the attempt to assess the potential benefits of an innovation in relation to its costs. Although efforts to encourage more analysis of costs and benefits go back twenty years at least, there are still very few examples of such work being done or influencing government policy.

Many reasons have been suggested for this lack, including the general inattention in education policy to economic analysis and a culture in schools that sees education as something intuitive, and therefore unlikely to benefit from more careful analysis. Yet it is clear that some educational innovations have much greater potential to produce desired outcomes, especially in relation to the widely varying costs of various innovations.

Although I have used the terms "costs" and "benefits", this analysis actually works from a cost-effectiveness framework, which assesses the effects, financial and otherwise, of various policy options in relation to the cost³. In this case, the alternatives are a) maintaining the present special school system for Romani children and b) integrating Romani children into mainstream classes and schools with appropriate supports.

In making a decision on this or any other policy issue, it is important to look not just at the costs, but also at the outcomes. Otherwise one runs the risk of adopting a policy that appears to be cheaper but produces worse results, or a policy that yields less benefit at higher cost. Around the world, integration of children into mainstream schools has been advocated not primarily on economic grounds, but for reasons of human rights and because of its short- and long-term benefits to children and society, so cost-effectiveness is an appropriate framework for analysis.

Benefits are usually considered as accruing to the individuals involved and to the broader community. Benefits to individuals could include greater self-esteem, higher levels of literacy, greater employability, higher income, better health, greater security and reduced stress, and so on. Benefits to the community are also broad. One area often examined is lower costs to the public for social services. More educated people are unemployed less, receive less social security, pay more taxes because they earn more, use less health care, and commit fewer crimes. All of this reduces costs to the public sector. At the same time, more educated people also tend to have greater participation in civic affairs and are able to contribute more to society in various intangible ways. Increased human dignity can also be seen as a benefit in its own right. All of these benefits are relevant, though many are hard to put into financial terms.

Issues in Conducting Cost-Effectiveness Studies

Cost-effectiveness analysis in education is not simple (B. Levin, 1999; Hummel-Rossi & Ashdown, 2001). The decision on what costs to include depends on having a theory of what resources and actions are needed to produce desired educational outcomes. Many educators see all of their resources as contributing to all outcomes, but in reality educational outcomes are jointly produced by a variety of inputs or programs, making it hard to disentangle effects of particular resources. For example, classroom teachers teach reading but parents and families are also vital in developing reading skills, while teachers in other subjects and support staff can also play an important role in shaping reading outcomes. Community facilities such as libraries can also play a significant role in promoting reading. If an analysis of reading costs included only classroom teachers it would understate the real resources quite significantly.

Costs are not always easy to determine (B. Levin, 1999). Some, such as teachers' salaries, are quite straightforward. Others, such as overheads (defined as those costs not directly part of programme delivery, such as administration or the operation of facilities), are more difficult to determine and to allocate to programs. In the case of specific interventions, one can usually calculate fairly readily the incremental costs. However if a new programme also changes the kinds of people who become involved, or the way that regular teachers do their work, there can be additional inputs that are not captured if one looks only at the incremental costs.

Moreover, if one sees education as being produced largely through the work of students (B. Levin, 1994), it is immediately apparent that a whole set of other resources are vitally important, even though they are not considered in most education cost analyses. Foremost among these are the efforts of students themselves and their families. If students increase their effort, or if parents increase their support for learning in effective ways, then student outcomes will normally improve, but there is no easy way to quantify the costs of these efforts. This is one reason why education systems do not pay more attention to ways of increasing student and parent effort even though such a strategy would appear to have very positive payoffs.

The assessment of educational outcomes is even more difficult. The desired outcomes for schools are diverse, long-term and hard to measure. The most commonly used outcome indicators, such as test scores or graduation rates, do not necessarily say much even about academic skills. At the same time, academic skills are much less important than long-term outcomes of education such as self-confidence, employment, earnings, or social participation. In the case of the Step by Step Roma Special Schools Initiative, the long-term ability of children to be productive citizens is probably the most important anticipated outcome. Yet these long-term results are very difficult to assess and even harder to attribute to specific interventions.

Related Literature

Two bodies of literature are relevant to this project. The first is work that discusses the costs of integrating students from special education settings into mainstream settings. This literature can shed light on the costs of the Initiative. The second body of work has to do with the longer-term benefits of improved educational outcomes for children, which bears on the longer-term benefits of the Initiative. The evaluation of the Roma Step by Step Special Schools Initiative gives good grounds for believing that integration, if well done, will indeed produce better outcomes for students. Particularly important is the finding that many Romani children are not actually mentally disabled and do not require the special education services that they are currently receiving.

Costs of Integration

Over the last 25 years, efforts have been made in a number of countries, including many in Europe, to reduce the numbers of students who are educated in segregated settings (European Agency, 1998). These efforts have been part of a growing commitment to understanding and respecting student diversity while still striving towards the best possible outcomes for all learners (Ainscow, 1999). The portion of students in segregated educational settings in Western Europe varies from country to country by as much as 400%. Nor has the situation improved in Europe in recent years (European Agency, 1998). However there is growing pressure to change this situation based on human rights arguments as well as educational considerations.

I was not able to locate any studies that deal with situations directly analogous to this Initiative – that is, looking at costs and benefits of mainstream-ing students who had mistakenly been placed in special programmes they did not require. However, the literature I did review does provide some useful guidance.

In considering the value of special education research it is important to keep in mind that most Romani children do not, in reality, have significant mental disabilities and therefore may not be in need of special education services. If, in fact, many Roma could be successful in mainstream classrooms, this is itself a powerful argument in favour of integration and also a reason to believe that integration could be achieved at a reasonable cost. (The conceptual and educational issues around Romani integration are not a focus of this article but are an important element in the larger Initiative.)

The costs of providing special education, regardless of setting, are typically at least double those of providing regular education (Chaikind et al., 1993). Provision of education in special classrooms and schools is even more expensive, and may cost up to six times as much as mainstream education (Organisation for Economic Cooperation and Development, 1999). This finding seems consistent in several countries even with different kinds of student needs and different definitions of costs. The difference in cost is primarily due to differences in staffing, since staff costs are by far the largest component of school expenditures. Special schools and classes tend to operate with more staff per student due to smaller classes and more support staff. They may also have additional overhead costs such as extra facilities or transportation.

The cost per student per year also hides the additional cost of students who take additional years to complete their programme. Requiring a student to repeat a year or grade doubles the cost of that student's progress, so programmes that permit students to make progress more quickly are automatically cost-effective. For example, studies have found that Reading Recovery, a New Zealand-based programme that works to improve the skills of poor readers in primary school, is expensive, but also shows good returns because it is successful in avoiding failure and grade repetition for many students (Cyer & Binkney, 1992).

Studies that compare segregated with integrated settings also find that the move to integration is cost-effective. For example, Pruslow (2001) compared three different models for providing special education services and found, using careful cost analysis, that inclusive models generated better outcomes in relation to costs.

A study by the Organisation for Economic Cooperation and Development (OECD) looked at practices in Australia, Canada, Denmark, Germany, Iceland, Italy, the UK and the USA found that:

…it seems safe to conclude that well developed inclusive practices, which give equivalent attention to disabled students, are less expensive than segregated ones. Furthermore, the evidence on educational progress is such to suggest tentatively that well structured and supported inclusive practices are beneficial for both disabled and non-disabled students alike. On this basis, inclusive provision would certainly seem to offer a cost-effective form of provision. (OECD 1999, p. 2)

Accordingly, the best available evidence would seem to suggest that integration is usually less expensive than segregated settings. As already noted, this is likely to be even more the case for Roma as the evidence suggests that many of the children in segregated special education settings are not actually mentally disabled and are quite capable of learning in mainstream settings with appropriate supports.

Short and Long-Term Benefits

The Step by Step Roma Special Schools Initiative found that Romani children taught the regular curriculum with appropriate supports will also have significantly better academic outcomes – improved levels of literacy and higher educational attainment (Rona & Lee, 2001). More students are expected to complete secondary schooling and to be eligible for post-secondary education.

Determining the impact of integrated and segregated placements on student outcomes is very difficult, because the two groups of students are not usually comparable. For example, some special schools address the needs of children with severe disabilities, whose life outcomes are likely to be worse than non-disabled students, regardless of the setting in which they are educated. Nonetheless, the consensus in special education literature is that more integrated settings are preferable, provided that students in such settings have adequate supports to achieve the best outcomes of which they are capable. Evidence from the Initiative indicates significantly improved educational outcomes for Romani children in the pilot mainstream classes.

More importantly for the purposes of this article, a very large body of evidence shows that improved educational outcomes for individuals, including increased self-esteem, higher levels of literacy and better educational credentials are associated with improved outcomes in almost every area of life. Vernez, Krop and Rydell (1999, p. 13) determined that education is a "public good whose benefits accrue not only to the individual attending school, but to society as a whole." Persons with more years of education or higher literacy consistently earn more income, pay more taxes, are less likely to be unemployed, commit fewer crimes, live longer and have fewer health problems (Vernez et al., 1999; Osberg, 1998). By increasing the education levels of the population at large or a segment of the population that is economically disadvantaged, governments could realise savings in social programmes. Higher levels of education are also associated with important benefits whose economic impacts are harder to measure, such as a greater levels of civic cohesion, increased volunteerism, greater propensity to vote, and higher levels of public trust (Dayton-Johnson, 2001). These findings are consistent across countries (OECD, 1997).

A World Bank report (2000) on the Eastern Europe and Central Asia region indicates that more years of education are associated with more years of work, higher earnings, and greater access to on-the-job training. For example, in both Hungary and the Czech Republic there is reliable evidence that the probability of unemployment declines steadily as literacy skills increase (2000, p. 23). Also, the likelihood of household poverty declines as education level increases. Work on the International Adult Literacy Survey (IALS) clearly shows that in every country, people with higher literacy levels had better earnings, better health, lower unemployment rates, and so on (Statistics Canada, 2001). Moreover, these gains clearly apply to people from disadvantaged settings even though their eventual outcomes may not totally catch up because of their weaker starting position (Lavin & Hyllegard, 1996).

Quantification of these benefits is difficult and subject to large variation depending on one's starting assumptions. However, many studies show significant long-term benefits to programmes that improve educational outcomes. Canadian research suggests that the lifetime cost to the public of a child that does not develop appropriate education and social skills can be as high as $1.5 million; clearly any measures that reduce the disability and required care will have large benefits. Mustard (2000) reviews many programmes from around the world that have aimed to improve educational outcomes for young children, and concludes that improving early education is a very powerful policy lever. Even programmes that support access to post-secondary education for under-educated adults appear to have good long-term returns both to individuals and to society (Lavin & Hyllegard, 1996; Alcorn & Levin, 2000). Vernez et al. (1998) estimate that increasing the educational attainment of native-born (i.e. not immigrants) men in the United States, would generate savings of 60 to 80 percent in lifetime public expenditure per person, more than covering the additional cost of providing the education.

These issues are especially important in Central and Eastern Europe, where, as the World Bank notes,

Just when human capital is becoming increasingly important in the region, inequities in learning opportunities are increasing – to the detriment of the poor everywhere (2000, p. v).

The Bank recommends in particular that countries try to improve educational outcomes for populations at risk, such as Roma (2000, p. 35).

There is also evidence that the success of minority groups in education is dependent on programmes that are sensitive to the group's culture and background (McDaniel, 1998). Thus a programme design that incorporates strong connections with families and communities and support for first languages is a recommended approach.

It can safely be concluded that a programme that allows children who have previously not succeeded in attaining levels of education permitting them access to gainful employment will produce substantial public and private returns on the investment in both economic and social terms, and that some additional expenditure to achieve such outcomes is well justified. All of this is, of course, in addition to the case for integration based on fundamental human rights.

Approach to Analysis of Costs

The Step by Step Roma Special Schools Initiative model involves measures aimed to support the moving of Romani children from special education settings, such as special schools or classes, to integrated settings in which they share their schooling with other children. A cost-effectiveness comparison should therefore assess the costs of each approach in comparison with the outcomes each approach yield.

The nature of schooling for Romani children and the alternatives available for integration vary from one setting to another both within and between countries. Given limits of time, I decided to examine a small number of cases in different kinds of settings on the assumption that a consistent pattern across the cases would likely apply more broadly in the countries as a whole.

Most of the Romani children in the Project were in one of two kinds of settings:

1. Special schools in which all children were regarded as mentally disabled (even though, as noted earlier, it is not evident that they were). The children were taught in a completely separate facility, geared entirely to children who are seen as unable to do mainstream school programming. They all followed a special curriculum with substantially reduced expectations. In some of these schools all, or almost all children were Romani, while in others, Romani children were only a part of the school population. Two options exist in this model:

• Move some children from the special school into a mainstream school and/or into regular classrooms in that school, with appropriate supports, while leaving the special school in place for those children still deemed to require its services;
   
• Change the special school into a mainstream school that teaches the mainstream curriculum and has the authority to grant mainstream diplomas. Such a school might continue to serve primarily Romani children who live nearby, but would provide them with a mainstream programme leading to success.

2. Remedial classes in mainstream schools, where children attend a mainstream school but spend all their time in a special class. These classes may be entirely, mainly, or partially made up of Romani children depending on the school. Integration in this model would mean moving the children into mainstream classes in the same school with appropriate supports.

It should be noted that integration is not just a matter of physical location. It also means that students would be studying the mainstream curriculum. Outcomes similar to those of other students would be expected of them.

Four schools were selected as cases so as to illustrate the various approaches to integration.

Assumptions

The study makes the following assumptions:

1. Any and all other specific supports for Romani children or their families, whether from the school or from other sources, would continue under integration. These items could include meals, housing, transportation, educational materials or family allowances.
    
2. Governments would not reduce the funding they provide to schools for Romani students as a result of integration, so there would be no incentive to keep students segregated.
    
3. Integration would be carried out in a way that did not penalise families whose children were integrated, so families would not have financial incentives to keep their children in special schools.

Since the costs of all these items would remain the same under each approach, they are not included in the analysis. However, governments need to be aware of these requirements.

Measuring Costs

The measurement of costs for this study is relatively straightforward. For ease of comparison, all costs were calculated in US dollars using 2001 conversion rates.

For each of the four cases, all the costs of current provision were calculated. These costs include teacher salaries, administrative and support staff salaries and operating costs of the school or programme, including feeding or boarding of students where this occurred.

The analysis assumed that these same overall financial resources would be available in the integrated approach as currently exists in the special school or class. It then asks whether the integrated approach could be supported with the same level of resources if they were used differently.

The costs of the integrated approach include its four main elements:

1. Teaching and support staff receiving Romani students must be given substantial additional training to allow them to work effectively with these students (and more effectively with all students).
    
2. Romani students must be in classes of reasonable size (preferably no more than 20 students) so that teachers have the time to work with them.
    
3. Romani teaching assistants or family coordinators (the terms are used interchangeably in this article) should be in place to work with students in the school and to provide a strong liaison between schools and families.
    
4. Teachers in mainstream classes should have access to education resource teachers to provide ongoing support.

The assumptions used about the various components are as follows:

Training

Costs were estimated for 180 hours (30 days) of training in the first year for each classroom teacher who has integrated children and for other staff as appropriate. Training includes the Step by Step method, school improvement and anti-bias education.

Calculation of training costs included both local and regional or national sessions including all costs of resource persons, travel, subsistence, substitute teachers for the half of the training that would occur during school time, and overhead costs.

On this basis, my estimate of training costs per teacher for 30 days of training ranged from $900 to $1800 with the difference due to the varying salary and cost levels among the countries.

In the second and third years, approximately 10 days of training would continue to be provided on a similar basis. This means that the first year training costs are high but are a one-time expenditure. High quality training is an investment with a long-term impact on teachers. Accordingly, the study calculated the three-year training cost and then amortises that cost over 10 years, which is a reasonable estimate of the lasting effect of training for teachers on average. Total three year costs per teacher ranged from $1600 to $3000, which, amortised over ten years, is from $160 to $300 per teacher per year.

Class Sizes

Class sizes vary substantially within and between schools, and the impact of adding additional students to a school depends on the precise circumstances in each school. For purposes of this analysis I assumed that any classes receiving Romani students would have an average size of 20 or less and be allocated additional salary costs on this assumption.

Salaries

Salaries for teachers, specialist teachers and classroom assistants were based on average pay rates supplied by Initiative staff in each country. These ranged from $1200 to $2500 annually for classroom assistants and from $1600 to $5000 annually for teachers and specialist teachers. Costs were lowest in Bulgaria and highest in the Czech Republic.

Romani Teaching Assistants/Family Coordinators

The study assumes that a number of Romani classroom or school assistants would be provided in schools receiving integrated Romani students. The optimal number of assistants in a school depends on the number of Romani children but would range from one to 3 or 4.

Special Education Resource Teachers

The model assumes that a position would be created in each integrated school for a person who is trained in working with children with diverse needs. This staff member would work primarily with teachers to help them develop appropriate techniques to recognise and respond effectively to the diverse needs of all learners.

Limitations

The data and conclusions in this study are drawn from a small number of cases and are based on assumptions in a pilot project. Although I believe these assumptions are reasonable and the conclusions are sound, it is possible that extension of this approach on a much wider scale could produce problems or issues that are not yet evident. For example, Iassumethat there is sufficient physical space in mainstream schools to take in additional children. The analysis also shows that there may be particular factors in each national setting that are also relevant. For example, in Hungary, the implications of the division of education funding between national and local levels would need to be considered.

Findings

General Framework

Both international research and data from the four countries indicates that the per pupil cost to educate a student in a special setting is typically at least twice as much as in a mainstream setting. This gives a strong intuitive case that integration would actually be less expensive even with additional supports. The question facing policy-makers is what configuration of resources – teachers, specialists, Romani teaching assistants, materials, and so on – will produce the best results.

Most special settings have very small class sizes (8-12 students) and a substantial number of full-time specialists such as speech therapists. Integrated students could be placed into mainstream classrooms with fewer students than is typical (about 20) but more students than are in each class in the special school. This would free staff positions and budget that could be redeployed to finance the additional costs of training, Romani assistants, and other supports.

For example, instead of having 8 classroom teachers and 4 specialists for 80 students, it would be possible for the same funds to have 5 classroom teachers, 1 specialist, 5 classroom assistants, and money for additional materials and training without any increase in budget.

The Four Cases

• Case 1 – A special school in a segregated village that serves about 120 students with 17 teachers at a total cost of approximately $100,000 annually.
• Case 2 – A special school that has 170 students and a staff of about 35, including 27 teachers, 3 Romani assistants and 4 specialists, with a budget of $160,000 per annum.
• Case 3 – A special school with 160 students in 13 classes. The total staff is 35, including 13 classroom teachers, 13 special education teachers and 8 support staff of various kinds. The school's total budget is about $180,000 per annum.
• Case 4 – A special school with approximately 300 children, of whom nearly half live in residence in the school. The staff totals almost 100 people including more than 50 teachers as well as specialists, administrators, boarding house staff and support staff. The total budget is about $200,000 per annum, of which half is for staffing of the school, about a quarter for operational and other costs and another quarter for operating the residence.

In each of the four cases, a careful analysis was conducted as to whether the Step by Step model could be financed by redeploying existing resources (see Table). The per-pupil operating costs of these schools are, in each case, about double the operating costs for regular schools in the same country, primarily because class sizes are very small and, in some settings there are a large number of additional professional staff. Redeploying these resources provides funds for maintaining class sizes at reasonable levels (less than 20), training for staff, Romani classroom assistants and additional curriculum materials.

Conservative assumptions were used throughout to ensure that the analysis did not have any bias in favour of integration. The analysis took into account the particular features of each setting, such as the financial contribution to education in Hungary by local authorities. The conservative assumption was made that the special school would remain open to serve students who were not being integrated and would retain at least the same per pupil funding as it currently has. For example, if 50% of the students were proposed to be redeployed to mainstream schools, the special school would retain more than 50% of its budget.

In each case it is clear that the integration of either some or all students into mainstream schools could occur, with appropriate supports, within existing resource levels, especially if one were to regard the initial investment in training as a cost to be amortised over several years. Integration of all students gives better financial results than integration of only some students, but both are affordable.

CA – classroom assistant
RCA – Roma classroom assistant

Note: All the proposed models provide all the elements of the model outlined earlier including staffing, training, resource teachers and additional materials. The cost of training is amortised over ten years.

In sum, cost does not appear to be a barrier to integration, even when providing an extensive package of supports as defined in the Step by Step Roma Special Schools Initiative.

Taking Benefits Into Account

I did not attempt to quantify the benefits of integration and the consequent improvement of outcomes for Romani children. However, if the results of the Step by Step Initiative are representative, the benefits are likely to be very substantial, both to the individuals involved and to the society as a whole. These benefits are both financial and social. Financial benefits would include higher lifetime earnings for individuals and lower state social security payments. Social benefits would include improved social cohesion. Again, these benefits are in addition to the advantages of integration in terms of human dignity and human rights.

In addition, the extensive training and support to teachers in integrated schools would improve the education of all the children in these schools, not only the Romani children being integrated. Teachers who are better trained will be more effective with all children. Smaller average class sizes will benefit all children. This would mean that for the same level of cost countries could see significant benefits in individual earnings, taxes paid, family stability, reduction of social security costs, lower health care costs, and so on for a substantial segment of the population.

Conclusion

The evidence reviewed in this study suggests that integration of Romani students is financially feasible and is likely to yield significant positive economic and social returns. This evidence supports arguments for integration based on a human rights rationale.

References

Ainscow, M. (1999). Understanding the development of inclusive schools. London: Falmer.

Alcorn, W. & Levin, B. (2000). Post-secondary education for indigenous peoples. Adult Learning, 11(1), 20-25.

Barnett, W. (1996). Lives in the balance: Age-27 benefits-cost analysis of the High/Scope Perry Preschool Program. Ypsilanti, MI: High/Scope Press. See also (www.highscope.org/research).

Barnett, W. & Escobar, C. (1987). The economics of early educational intervention: Areview. Review of Educational Research, 57(4), pp. 387-414.

Chaikind, S. et al. (1993). What do we know about the costs of special education? A selected review. Journal of Special Education, 26(4), pp. 344-379. (www.ericae.net/ericdb/EJ459618.htm).

Dayton-Johnson, J. (2001). Social cohesion and economic prosperity. Toronto: Lorimer.

Dyer, P. & Binkney, R. (1992). Reading recovery: A cost-effectiveness and educational-outcomes analysis. ERS Spectrum, 10, pp. 10-19.

European Agency for Development in Special Needs Education (1998). Integration in Europe: Provision for pupils with special educational needs. Copenhagen: EADSNE. (www.european-agency.org/publications).

Hummel-Rossi, B. & Ashdown, J. (2001). The state of cost-benefit and cost-effectiveness analyses in education. Review of Educational Research 72(1), 1-30.

Karoly, A.; Greenwood, P. Everingham, S., Hoube, J., Kilburn, M., Rudell, C., Sanders, M. & Chiesa, J. (1998). Investing in our children. Santa Monica, CA: RAND.

Lavin, D.E. & Hyllegard, D. (1996). Changing the odds: Open admissions and the life chances of the disadvantaged. New Haven, CT: Yale University Press.

Levin, B. (1994). Improving educational productivity through a focus on learners. Studies in Educational Administration, 60, pp. 15-21.

Levin, B. (1999). Value for money in the National Literacy/Numeracy Strategies Unpublished report to the Department for Education and Employment (Britain). November.

Levin, H. & McEwan, P. (2001). Cost-effectiveness analysis, 2nd edition. Thousand Oaks, CA: Sage.

McDaniel, S. (1998). Toward healthy families. In Canada Health Action, Volume 3: Determinants of Health: Settings and Issues. pp. 7-46. Sainte-Foy, Quebec: MultiMondes.

Mustard, F. (2000). Early child development and the brain. Unpublished paper prepared for the World Bank.

OECD, 1997. Labour market policies: New challenges – Lifelong learning to maintain employability. Paris, France: Organisation for Economic Co-operation and Development.

OECD (1999). Sustaining inclusive education: Including students with special educational needs in mainstream schools. Paris: OECD (CERI/CD(99)2).

Osberg, L. (1998). Economic policy variables and population health. InCanada Health Action, Volume 3: Determinants of Health: Settings and Issues. pp. 579-610. Sainte-Foy, Quebec: MultiMondes.

Pruslow, J. (2001). A comparative cost and utility analysis of three models of service delivery for special needs students. Journal of Education Finance, 27, pp. 535-566.

Rice, J.K. (1997). Cost analysis in education. Educational Evaluation and Policy Analysis, 19(4), pp. 309-317.

Rona, S. & L. Lee (2001). School Success for Roma Children: Step by Step Special Schools Initiative, Report. New York, NY: Open Society Institute, 2001.

Statistics Canada (2001). Literacy, numeracy and labour market outcomes in Canada. Ottawa: Statistics Canada. (www.statcan.ca)

Vernez, G., Krop, R. & Rydell, P. (1999). Closing the education gap: Benefits and costs. Santa Monica: RAND. (www.rand.org).

World Bank (2000). Hidden challenges to education systems in transition economics.

Washington: World Bank, Europe and Central Asia Region, Human Development Sector.

Endnotes:

1. Benjamin Levin is Professor of Educational Administration at the University of Manitoba and was, until recently, chief civil servant for education for the Province of Manitoba, Canada. This study was originally prepared for the Open Society Institute, whose support is gratefully acknowledged. Liz Lorant, Susan Rona and Linda Lee were particularly helpful. I also thank the country coordinators of the Step by Step programme of the Open Society Institute, who provided the data and reviewed my findings. All opinions and any errors are the sole responsibility of the author.
2. Integration here refers to placing students in standard school programmes, where they previously have been educated in schools or classes for children with disabilities or other substandard arrangements. This article does not deal with geographic or residential segregation.
3. Cost-benefit analysis, strictly speaking, requires all benefits to be put into financial terms, which is not a realistic possibility in the frame of this study.