Horváth and Kiss v. Hungary: a strong new Roma school segregation case

The Strasbourg Court has once more delivered a judgment in a Roma school segregation case. The applicants in Horváth and Kiss v. Hungary are two young Roma men, who were diagnosed as having mild mental disabilities when they were children. As a result of these diagnoses, they were placed in a remedial school. Their education there was poor: the curriculum was underdeveloped, their schooling did not give them access to the type of job they wanted, and they ended up de facto segregated from the wider population. The applicants claim that their education in the remedial school constituted ethnic discrimination in their enjoyment of their right to education (Article 2 Protocol 1 in conjunction with Article 14 ECHR). The Strasbourg Court rules that they’re right: it finds a violation of the Convention on the ground of indirect ethnic discrimination.

Much of the reasoning in this case is familiar from other Roma school segregation cases, such as, notably, D.H. and Others v. Czech Republic (2007) and Oršuš and Others v. Croatia (2010). In this post, I will discuss the judgment and try to highlight what’s new in the Court’s reasoning.


I’ll be brief about the kind of steps the Court takes in an indirect discrimination case such as this one: the main principles are familiar from the other Roma school segregation cases. In a nutshell, then, the Court first establishes that there is a prima facie case of discrimination, because both in the past and in the present Roma kids have been overrepresented in remedial schools in Hungary in general, and also in this one in  the city of Nyíregyháza  in particular (par. 110-111). It then falls to the State to prove that “the difference in treatment had no disproportionately prejudicial effects”(par. 112). At this point, the judgment becomes a bit rambling. For the Court, the main issue seems to be:

to ascertain to what extent special safeguards were applied that would have allowed the authorities to take into consideration, in the placement and regular biannual review process, the particularities and special characteristics of the Roma applicants who sat them, in view of the high risk of discriminatory misdiagnosis and misplacement. (par. 121)

 The answer to this is “no”; the State did not put in place the necessary safeguards to prevent misdiagnosis. The Government itself had acknowledged that the tests, which were the same for Roma children as for children from the ethnic majority, were partly culturally biased (par. 120). It is for this reason that the Court finds that the State’s argument that it properly tested the Roma kids, cannot serve as a sufficient justification for the impugned treatment (par. 123). The Court concludes that:

As a consequence, [the applicants] received an education which did not offer the necessary guarantees stemming from the positive obligations of the State to undo a history of racial segregation in special schools. The education provided might have compounded their difficulties and compromised their subsequent personal development instead of helping them to integrate into the ordinary schools and develop the skills that would facilitate life among the majority population. (par. 127)


First of all, what I appreciate about this judgment is that the Court discusses both the problems relating to the long-standing inequality and discrimination of Roma, and the difficulties of children with learning disabilities. You might say that the Court takes an intersectional approach; combining a Roma-equality perspective with a mental disability perspective. The Court’s reasoning captures well the complexity of the situation wherein the applicants find themselves.

The other thing I appreciate is that the Court does not go along with the argument of the applicants that “Roma were uniquely burdened by the current system; no other protected group had been shown to have suffered wrongful placement in special schools based on the diagnostic system.” (par. 91). I can see why the applicants would make such an argument: they thought that in order to establish a prima facie case of discrimination, they had to prove that – as a group – they had it worst. Fortunately, the Court holds that a vulnerability competition (meaning a competition amongst groups for recognition of their vulnerability) is not necessary. The Court namely observes that:

a general policy or measure exerted a disproportionately prejudicial effect on the Roma, a particularly vulnerable group. For the Court, this disproportionate effect is noticeable even if the policy or the testing in question may have similar effect on other socially disadvantaged groups as well. The Court cannot accept the applicants’ argument that the different treatment as such resulted from a de facto situation that affected only the Roma (par. 110).

The lesson here for applicants and their lawyers is that they need not enter into a vulnerability competition in order to have their indirect discrimination claims recognized. In my view, such a competition could only have perverse effects; a race to the bottom is not what Strasbourg case law should be about.

So what’s new?

I think the novelty of this judgment lies in the Court’s reasoning about positive obligations. To my knowledge this is the first time that the Court has talked of “the positive obligations of the State to undo a history of racial segregation in special schools” (par. 127). “[I]n light of the recognised bias in past placement procedures”, the Court says “that the State has specific positive obligations to avoid the perpetuation of past discrimination or discriminative practices disguised in allegedly neutral tests.” (par. 116).

So far, the kind of positive obligation that the Court recognized in Roma school segregation cases had been procedural. In Oršuš, for example, the Court spoke of the obligation to put in place “safeguards that would ensure that . . .the State had sufficient regard to [Roma children’s] special needs as members of a disadvantaged group” (Oršuš, par. 183). In the present case, the Court goes a step further by insisting on a substantive positive obligation, namely to “undo a history of racial segregation in special schools”.

This goes quite far. But whether this entails a specific obligation on the State to provide specialized tests for children from a Roma background, and for children from other socially disadvantaged backgrounds, I am still not sure.

9 thoughts on “Horváth and Kiss v. Hungary: a strong new Roma school segregation case

  1. Dear Alexandra,
    I am a great fan and particularly grateful this time for your comments on Horváth and Kiss.
    I agree with much of your analysis and would like to offer further insight into this case and how it was in fact argued. Regrettably, the judgment does not fully reproduce the applicants’s arguments, but similar to DH and Others, this case was argued primarily as a direct race discirmination case, and only secondarily as an indirect race discrimination case. The submission and applicants’ observations can be downloaded from the bottom of this page: http://www.cfcf.hu/nyiregyhaza-felrediagnosztizalas_hu.html
    The case grew out of a strategic litigation effort initiated by then MEP Viktória Mohácsi. The clients were hand-picked by experts out of 60 children. In 2005, five sets of cases were taken to civil courts in Hungary and the only one that succeeded at least at one domestic instance was Horváth and Kiss. The Hungarian Roma rights movement sought to establish much of what the ECtHR found in this case.
    1. misdiagnosing otherwise mentally sound Roma children as mentally disabled amounts to race (ethnicity) discrimination.
    2. basically no justification exists for the above race (ethnicity) discrimination, because race discrimination is subject to strict scrutiny.
    3. in the Hungarian context, misdiagnosis results in segregation, whereby children such as the applicants study in special schools. In Hungary, non-disabled children study in mainstream (normal) schools, whereas disabled (meaning physical, sensory and mental disability) children study in special schools. (There are concessions made in the system, but let me keep the description simple.) Special schools are thus segregated on the basis of disability and their curriculum is lower. This stigmatises disabled children on the ground of their disability and stigmatises misdiagnosed (meaning: non-disabled) Roma children on the ground of their race. The two-tier system of public education exists in the whole CEE region, hence the analogies with DH and Others. Significantly, the IQ scores used to determine the upper limits of mental disability in HU at the material time were higher than those proposed by WHO (para.118), which made misdiagnosis even more likely.
    4. States Parties are under the duty to remedy race discrimination (misdiagnosis) by way of implementing positive action measures.
    5. what was achieved in DH and Others at the appeal level with a divided vote was consolidated in Horváth and Kiss with a unanimous vote
    6. the bonus finding is obiter dicta: discrimination on the ground of mental disability will also be subject to strict scrutiny and it may be that a state duty exists to remedy discrimination based on this ground as well (para. 128.).
    Misdiagnosis = direct race discrimination
    Significantly, the case was argued on the basis of EU anti-discrimination law transposed into domestic law (namely the Racial Equality Directive, 43/2000/EC, RED). The applicants construed misdiagnosis as direct race discrimination. Under the RED and HU law misdiagnosis amounts to direct discrimination, because 1. the applicants had a protected ground (race, ethnicity), 2. they suffered less favourable treatment on this ground (on the ground of their Romani ethnic origin they were treated as mentally disabled), and 3. their comparators were treated more favourably (ethnic majority children were not misdiagnosed). The burden of proof provisions in HU required the applicants to establish their protected ground and the less favourable treatment, but not the causal link between the protected ground (race, ethnicity) and the less favourable treatment (being treated as mentally disabled even though they were NOT mentally disabled).
    In essence, the applicants argued that the diagnosis and the IQ tests used for determining which type of school a child shall attend were racially biased. In order to support this claim, they relied on the 40 year long and well documented history of the misdiagnosis of Roma children in HU, the deviation from WHO standards and the statistical data that showed a gross overrepresentation of Romani children in special schools. Last, they relied on the Court’s assessment in DH – which was repeated in Horváth and Kiss as well, para. 121 and 123 – that the tests in question were in fact culturally (meaning: racially) biased.
    Secondarily, the applicants invoked the CJEU’s Maruko judgment, arguing that the diagnostic system has for 40 years been proven to disadvantage one group and one group only, that of the Roma. The Court rejected this scientific data based argument in a rather summary fashion, without even requiring the respondent State or itself to provide a reason why and based on what it had done so (para. 110).
    Misdiagnosis is not intersectional (multiple) discrimination
    As you may have noticed above, what causes confusion in misdiagnosis cases is that the less favourable treatment of non-disabled Romani children roughly equals the less favourable treatment disabled children (of any ethnic origin: Romani, majority, minority, etc.) suffer in special schools. In other words, race discrimination is built on disability discrimination. The segregation of disabled children in special schools is the conditio sine qua non of the segregation of misdiagnosed Roma children in the same schools. If disabled children had not been segregated in special schools but taught in integrated mainstream schools, misdiagnosed Roma children could not be segregated in special schools either.
    This causal link may be the reason why the Court felt compelled to call attention to prejudice against the mentally disabled (para. 128.).
    However, misdiagnosis can NOT successfully be argued on an intersectional (multiple) ground. Why? Because in an intersectional case discrimination is based on the victims’ two or three or more protected grounds taken together. The comparator must then be construed in a manner that she is different in multiple grounds: non-disabled Roma child compared to disabled non-Roma child, etc. Obviously, a Roma child who claims that she was misdiagnosed does not have more than one protected ground. This protected ground is her race/ethnicity. She is not in fact disabled. Disability comes into her case, because she is treated as if she was disabled. This constitutes her less favourable treatment: (even though she is not disabled) she is treated as being disabled.
    It is easy to see that if she based her claim on race and disability taken together, her whole case would collapse. Once a protected ground equals the less favourable treatment, there is in fact either no protected ground or the treatment is not less favourable. If all disabled children are referred to special school, it is irrelevant whether or not they are Roma or of majority ethnic origin.
    Indirect discrimination? Really?
    I believe that the ECtHR construction of indirect discrimination in this case – as well as in the other Roma education cases – collapses the difference between direct and indirect discrimination as understood under RED Article 2. Once the Court suggests that the apparently neutral criterion or practice (meaning: IQ tests or diagnosis) runs at least the risk of being culturally biased (para 121 and 123.), then it essentially admits that the impugned practice is biased against the Roma. In other words, it is not neutral viz. the Roma. This is the key finding, because apparently neutral means that the impugned practice shall be apparently neutral viz. the protected ground. If tests run the risk of being culturally biased, misdiagnosis cannot in fact be construed as indirect race discrimination.
    The bottom line is that in none of the Roma education cases has the Court found the justification offered by respondent States plausible. Given that in Orsus the justification relied on ethnicity based positive action measures – the only justification allowed under RED Article 5 – it will be exciting to see what if any justification the Court will allow.
    Against this backdrop, one may think that the only reason why indirect discrimination was found is political – because this form of discrimination is seen as NOT intentional. In practice, howeevr, it is the essence of judgments and not the labels attached to the kind of discrimination that matter. In essence, I am happy with the essence.
    Lilla Farkas
    counsel for Horváth and Kiss

  2. Dear Alexandra,

    There is one more point you raised and I forgot to reflect upon.
    It seems legitimate to me that you wonder what sort of positive action measures may the Court have had in mind. That I certainly do not know. What I know is what sort of positive action measures the applicants had in mind.
    The applicants in their submissions relied extensively on the Roma Education Fund publication, Pitfalls and Bias (available here http://www.romaeducationfund.hu/sites/default/files/publications/pitfalls-and-bias-screen_singlepages.pdf).
    This policy paper identifies two main types of positive action measures that can curtail misdiagnosis: 1. if the special school – (normal) mainstream school divide stays in place, reassessment must be regular and the tests need to be reformed to be less culturally biased, 2. if the government is prepared to shut down special schools – and thus stop the segregation of all disabled children as well as misdiagnosed Roma children – then individual assessment must serve the purpose of finding out what needs students have.
    REF is also recommending that while a decision is taken, entry testing should be suspended altogether.

    Thank you for your reference to the above Turkish case. I agree that on occasion the Court fails to distinguish properly among the different types of discrimination. What bothers me more is when it fails to see discrimination and fails to notice that if there is prejudice, then there is discrimination – even if ‘scientific’ facts challenged in the case seem correct. In Aksu v Turkey, for instance, I wondered what was scientific about an academic’s collection of widely held biased views from the ranks of the police forces. Other than these biased hearsay sources, were there any other ones to support the statement that Roma were an ethnic group closely tied to criminality? And even if there had been, would it be right to allow the spreading of racial prejudice in academic publications subsidised by the state?

    Coming from a country where the extreme right campaigns against the Roma as if they were all criminals and rallies in Gypsy settlements shouting ‘There is Gypsy crime’, I very much doubt that it is within the limits of free speech for an academic publication subsidised by the state to promote the same racist idea. The said academic is always free to publish from his own resources and under his own name whatever nonsense he wishes to put on paper.


    • Dear Ms. Farkas,

      Thank you for your insights on the litigation of the case Horváth and Kiss.

      In your last comment, you detailed what kind of positive action measures the applicants had in mind when they presented their case to the ECHR. I do not understand why the applicants did not make specific requests in terms of remedies. There was an opportunity for asking for systemic or transformative changes, but the applicants only claimed compensations for the costs and expenses incurred before the Court.

      I am trying to identify how the ECHR takes into account systemic discrimination, and its role in remedying it, but so far, my findings are poor, and I have the impression that applicants could play a greater role in asking for systemic changes.

      I would appreciate your comments in this regard.

      Best regards,
      Anne-Claire Gayet
      Lawyer doing research on systemic discrimination

  3. […] This Chamber judgment dealt with the issue of de facto segregation of Roma children in Hungarian schools, as a result of biased intelligence tests misleadingly suggesting the children had mental disabilities. The Chamber judgment is nominated for best judgment because, in finding a violation of article 14 juncto article 2 of Protocol 1 ECHR, the Court clarifies that States are under a positive obligation to undo a history of racial segregation in special schools. For more, see our post here. […]

  4. […] The segregation of Roma children in public education continues to be a major failure of the European human rights regime. In 2007 in D.H. v. the Czech Republic (also known as the Ostrava case) the ECtHR established the basic premises of challenging segregation of Roma children in public education. Importantly, the ECtHR agreed to accept statistical evidence to ascertain a violation, to reverse the burden of proof and also insisted on the burden of justification being as strict as possible for discrimination based on nationality or ethnicity. Despite being a seminal decision, however, D.H. v the Czech Republic is still awaiting its enforcement. The Chamber judgment in Horváth and Kiss v Hungary reinforces the line of jurisprudence marked by D.H. and subsequent decisions.  It confirms the admissibility of statistical evidence for establishing prima facie discrimination based on ethnicity (para 107) and the reversal of the burden of proof (para 108). It also reaffirms that in a public education setting it is not necessary to prove discriminatory intent for indirect discrimination (para 106). Importantly, the Court established that in the context of public education, in a case where a group of pupils has suffered past discrimination with continuing effects, structural disadvantage needs to be addressed by positive measures. The Court insisted on “particularly stringent” positive obligations due to the actual history of discrimination in the case (para 104). It is in this latter respect that the new Hungarian cases added significant insight to the existing jurisprudence. Applicants in the Hungarian case challenged the misdiagnosis of Roma children as mentally disabled, and their subsequent placement in segregated special schools. The curriculum in these special schools is more rudimentary than in ordinary public schools and educational opportunities for graduates of special schools are limited. Although placement to special schools is based on complex testing, the overall pattern still was that Roma children have been considerably overrepresented in special schools, compared to the proportion of Roma in the general population. The special school in the applicants’ town had been 40-50 per cent Roma, while Roma children amounted to 8.7 per cent of the student body in the same town (see para 4). The case before the ECtHR arose from a strategic litigation effort by Roma rights NGOs, the Chance for Children Foundation (CFCF) and the European Roma Rights Center. In order to litigate the underlying claims across the Hungarian judicial system, CFCF first of all had to establish the misdiagnosis of the applicants as students with mental disabilities. To this effect counsel for the applicants did not only furnish evidence on the racial bias implicit in the various tests used for placement in special schools, they also presented the results of the alternative testing of the applicants at a summer camp which yielded significantly higher test scores than those on the same tests which were conducted by government experts before (paras 31-34). The summer camp was funded by the Roma Education Fund, which later also submitted an amicus curiae brief in the domestic proceedings. These findings did not come as a surprise. The misdiagnosis of Roma children, who were placed the special schools for children with learning difficulties as a result of ethnic bias in the testing is a systemic and lasting problem in Hungary, as has been clearly exposed on the European level before. Unlike in many other countries in the region, in Hungary reliable data on the misdiagnosis of Roma children were available since the 1970’s. CFCF heavily relied on these data during the entire procedure. The findings of extensive European monitoring in several reports and recommendations on Hungary, as identified by the Advisory Committee on the Framework Convention for the Protection of National Minorities, the CoE Commissioner for Human Rights and ECRI’s regular monitoring, were consulted by the ECtHR in the case. The domestic procedure was complicated by numerous changes in the legal regulation of public education and also in the testing regiment itself. These changes were triggered in parts by the government’s alleged own efforts to improve testing. Also, the applicable Hungarian legal regulation was amended with the transposition of EU equal treatment rules alongside the introduction of a comprehensive equal treatment act and the establishment of a national equal treatment authority. The resulting procedure was rather complex with three respondents (the panel of experts misdiagnosing the applicants, the special school which the applicants attended and the county council which was responsible under the law for the operation of the panel of experts and the specials school) being brought to court, on appeal and also on judicial review in a discrimination suit with claims for damages. The applicants were not completely unsuccessful with their claims before domestic court, a factor which prompted a very close inquiry into the exact scope of their claim before the ECtHR during the preliminary analysis. After all, the panel of experts was found to act in a discriminatory fashion already by the first instance court for failing to assess the applicants in an individualized manner, and in the review proceedings the Supreme Court found that the county council had to pay damages for its failure to supervise the legality of the operation of the panel of experts. At the same time, the Supreme Court found no equal treatment violation by the special school and the county council, noting, that the failure of the state to remedy a systemic violation of human rights which resulted from the absence of a professional protocol for testing was a problem to be remedied by the ECtHR or by the Hungarian Supreme Court. It was in respect to this discrimination challenge contesting a systemic violation that the ECtHR found the case admissible (para 86). The ECtHR, however, refused to address the part of the petition which challenged the unsuitability of the tests used, as it did not find domestic remedies exhausted in this respect (para 87). As a result the ECtHR examined whether committing the applicants to otherwise segregated special schools for students with mental disability due to their Roma origin amounted to indirect discrimination with regard to the applicants’ right to education (Article 2 of Protocol no. 1 in conjunctions with Article 14). Indeed, the most significant novelty the Horváth and Kiss judgment brings is the Court’s analysis on the nature and extent of the state’s positive obligation to prevent a well-established, lasting and systemic violation of human rights in the context of education. The Court said explicitly that “the State has specific positive obligations to avoid the perpetuation of past discrimination or discriminative practices disguised in allegedly neutral tests” (para 116). It is for the state to demonstrate that the tests used as well as their application in practice is capable of “fairly and objectively” determining the learning abilities of the applicants (para 117). In the application of this standard, it turned out to be rather demanding. Importantly, before the ECtHR the government did not dispute the racial bias in at least some of the tests used, instead, it argued that “cultural bias could be compensated” by alternative examination (paras 95 and 120). The government, nonetheless, insisted that the over-representation of Roma children in the special schools results from social deprivation, a factor which is outside the scope of the right to education (para 96). Responding to the government’s points, the ECtHR found that exactly for these reasons there was at least a “danger” that the tests were culturally biased, therefore, the Court was looking for “special safeguards” to prevent misdiagnosis (para 121), which it did not find. The lack of safeguards was established with the ECtHR relying on the facts as established by the domestic courts, taking into account the rapidly changing legal environment which was also pointed out by the national courts. Note that while the ECtHR adopted the language of the government’s explanation, classifying racial bias as cultural bias, it did not follow the government all the way to accepting the social deprivation argument as an explanation for the disproportionately high number of Roma children in special schools in Hungary. Indeed, the Court relying also on reports from European monitoring bodies revisited the impact of the dubious concept of “familiar disability” used to justify the misdiagnosis of Roma children (para 115). This is a most welcome development, as the ECtHR appears to be willing to come to an informed assessment of developments on the ground, taking advantage of a wide range of sources for its analysis. In addition to the Court’s openness to statistical evidence in order to establish indirect discrimination on the basis of ethnicity, it is reassuring to see that the Court is open to listen and learn from the findings of extensive European human rights monitoring and reporting on questions which are too complex (and costly) for applicants to explore on their own in individual cases. The openness of the ECtHR to accept a wider range of evidence, and piece together a truly European account of a structural and systemic violation is all the more significant, as similar data on ethnic origin are not available in many other countries of the affected region. In closing, it is worth noting that –as also signaled by the ECtHR– special schools for mentally disabled and developmentally challenged children present a human rights problem in themselves. The Hungarian special school regime is particularly problematic, as it caters to children with mental disabilities, and also with other learning difficulties. Departing from the WHO value of IQ 70, pupils at IQ 86 or below were placed in these special schools. In this regime pupils with IQ 70 to 85 were understood to have a borderline intellect, and as such they were seen to have serious and persistent learning impairments. While the government requested in 2004 to stop the transfer of children with a score above IQ 70 to special schools, in 2007 the National Expert and Rehabilitation Committee still insisted on the borderline qualities of these children. In addition, as the Court also noted, in 2007/08 only 0.4-0.6 per cent of children with special needs were integrated in Hungary in secondary education (para 8), i.e. the overwhelming majority of special schools segregate their pupils from mainstream public education. The ECtHR accepted the government’s position that it maintained special schools  “to find a solution of children with special education needs” (para 113). At the same time, the Court expressed serious concerns about the existing regime, echoing the concerns of other European institutions (para 113).  In light of this background, it is not a surprise that the Court used the closing paragraphs of the judgment as an opportunity to emphasize (reinforcing its decisions in Kiss v Hungary) that due to the special vulnerability of persons with mental disabilities the limitation of their fundamental rights has to be justified by very weighty reasons, and has to be subject to strict scrutiny (Kiss, paras 42 and 44, reaffirmed at Horvath and Kiss, para 128). This language of clear encouragement from the Court should be read by disability advocates as a sign that the ECtHR may be ready to hear from them on the segregation of children with learning difficulties in several European countries. See also other comments on the case here. […]

Leave a Reply