6.3. Role of statistics and other data – Handbook on European non-discrimination law

Last Updated on August 11, 2019 by LawEuro

Handbook on European non-discrimination lawContents

Statistical data can play an important role in helping a claimant give rise to a presumption of discrimination. It is particularly useful in proving indirect discrimination, because in these situations, the rules or practices in question are neutral on the surface. Where this is case, it is necessary to focus on the effects of the rules or practices to show that they are disproportionately unfavourable to specific groups of persons by comparison to others in a similar situation. The production of statistical data works together with the shift of the burden of proof: where data shows, for example, that women or disabled persons are particularly disadvantaged, it will be for the state to give a convincing alternative explanation of the figures. The ECtHR spelt this out in the case of Hoogendijk v. the Netherlands:

“[T]he Court considers that where an applicant is able to show, on the basis of undisputed official statistics, the existence of a prima facie indi- cation that a specific rule – although formulated in a neutral manner – in fact affects a clearly higher percentage of women than men, it is for the respondent Government to show that this is the result of objective factors unrelated to any discrimination on grounds of sex.”[681]

When considering statistical evidence, the courts do not appear to have laid down any strict threshold requirement that needs to be evidenced in establishing that indirect discrimination has taken place. The CJEU does emphasise that a substantial figure needs to be achieved. A summary of CJEU case law is presented in the Opinion of Léger AG in the Nolte case, where he stated in relation to sex discrimination:

“[I]n order to be presumed discriminatory, the measure must affect “a far greater number of women than men” [Rinner-Kühn[682]] or “a considerably lower percentage of men than women” [Nimz,[683] Kowalska[684]] or “far more women than men” [De Weerd[685]].

Cases suggest that the proportion of women affected by the measure must be particularly marked. In Rinner-Kühn, the Court inferred the existence of a discriminatory situation where the percentage of women was 89 %. In this instance, per se the figure of 60 % […] would therefore probably be quite insufficient to infer the existence of discrimination.”[686]

Accordingly, when assessing statistics, national courts have to determine if they cover enough individuals to exclude fortuity and short-term developments.[687]

Example: A case[688] from Denmark concerns dismissals made in a government agency due to the need to reduce the workforce. All of the dismissed employees were above 50 years of age. The two complainants claimed that they had been discriminated against because of their age. The Supreme Court stated that statistical information could establish an assumption for discrimination because of age. However, the court found that a number of employees in the government agency who were older than the claimants had not been dismissed during the process of reducing the workforce. On this basis, the court concluded that, in this case, the statistical data regarding the age of the dismissed employees, as well as information about the age composition in the government agency, did not establish any facts which amounted to possible discrimination.

Example: In Hilde Schönheit v. Stadt Frankfurt am Main and Silvia Becker v. Land Hessen,[689] a part-time employee alleged that she was discriminated against on the basis of her sex. The difference in payable pensions, which was not based on differences in the time worked, meant that part-time employees were, effectively, paid less than full-time employees. Statistical evidence was brought to show that 87.9 % of part-time employees were women. As the measure, although neutral, negatively affected women disproportionately to men, the CJEU accepted that it gave rise to a presumption of indirect discrimination on the basis of sex. Similarly, a disadvantage to part-time workers, where 87 % of these were women was accepted as sufficient in the Gerster case.[690]

Example: In Lourdes Cachaldora Fernández v. Instituto Nacional de la Seguridad Social (INSS) and Tesorería General de la Seguridad Social (TGSS),[691] the claimant had paid contributions to the Spanish social security scheme for almost forty years. During that period, she had mostly been engaged in full- time employment, except between 1998 and 2005, when she had first been employed part-time and had then been unemployed. In 2010, she had applied for invalidity pension. According to the relevant law, invalidity pension was calculated on the basis of a period of eight years prior to the occurrence of the event giving rise to the invalidity. Workers who had engaged in part- time work during a period immediately preceding a period of unemployment were granted a reduced invalidity pension. The reduction came about as a result of applying the part-time work coefficient. Consequently, through this method of calculation, the claimant’s invalidity pension had been significantly reduced. The referring court had asked whether the relevant national provision could have been considered as discriminatory towards workers who had engaged in part-time work during a period immediately prior to an interruption of their contributions to the Spanish social security scheme. It had referred to the fact that, given that there are far more female part-time workers in Spain than male part-time workers, women would be particularly affected by this provision. The CJEU noted, however, that these provisions were not applicable to all part-time workers. They applied only to a limited group of workers, including the claimant, who, after a period of part-time employment had a gap in their contributions during the reference period of eight years. Consequently, global statistical data concerning part- time workers taken as a whole were not relevant when establishing whether or not women are more affected by the provisions of Spanish law than men.

Example: The Seymour-Smith case[692] concerns UK law relating to unfair dismissal, which gave special protection to those who had been working for longer than two years continuously with the particular employer. The complainant alleged that this amounted to indirect discrimination based on sex, since women were less likely than men to satisfy this criterion. This case is interesting because the CJEU suggested that a lower level of disproportion could still prove indirect discrimination “if it revealed a persistent and relatively constant disparity over a long period between men and women”. However, on the particular facts of this case, the CJEU indicated that the statistics that were presented, which indicated that 77.4 % of men and 68.9 % of women fulfilled the criterion, did not prove that a considerably smaller percentage of women could comply with the rule.

A similar approach can be found in the jurisprudence of the ECtHR.

Example: In Di Trizio v. Switzerland,[693] the applicant, who had been working full-time, was obliged to stop working due to back pain. She was granted a disability allowance which was discontinued after she gave birth. The competent authorities based the decision regarding her entitlement to the allowance on the ‘combined’ method. They had assumed that, even without her disability, she would not have been employed full-time after the birth of her children. The ECtHR noted that the applicant would probably have received partial disability allowance if she had worked full time or had devoted her time entirely to her household. Furthermore, it relied on statistics proving that 97 % of persons affected by the combined method were women who wished to reduce their working hours after birth of a child. Consequently, the statistics provided sufficient reliable information to establish a presumption of indirect discrimination.

Example: The case of D.H. and Others v. the Czech Republic[694] involved complaints by Roma applicants that their children were excluded from the mainstream education system and placed in ‘special’ schools intended for those with learning difficulties, on the basis of their Roma ethnicity. The allocation of Roma children to ‘special’ schools was based on the use of tests designed to test intellectual capacity. Despite this apparently ‘neutral’ practice, the nature of the tests made it inherently more difficult for Roma children to achieve a satisfactory result and enter the mainstream education system. The ECtHR found this to be proved by reference to statistical evidence showing the particularly high proportion of pupils of Roma origin placed in ‘special’ schools. The data submitted by the applicants relating to their particular geographical region suggested that 50 to 56 % of special school pupils were Roma, while they only represented around 2 % of the total population in education. Data taken from inter-governmental sources suggested between 50 to 90 % of Roma attended special schools in the country as a whole. The ECtHR found that while the data was not exact it did reveal that the number of Roma children affected as ‘disproportionately high’ relative to their composition of the population as a whole.[695]

Example: In Abdu v. Bulgaria,[696] the applicant and his friend, both Sudanese nationals, had been involved in a fight with two Bulgarian youths. According to the applicant, they had been attacked by the two young men who had verbally insulted them with racist remarks. The proceedings against the attackers were discontinued on the basis that it was not possible to ascertain who had initiated the fight and their motives. The authorities had not questioned the witnesses and had not interrogated the alleged attackers about the possible racist motive of their actions. The ECtHR found that the authorities had been in possession of evidence of a possible racist motive, and they had failed to conduct an effective investigation into it. In its judgment, the ECtHR referred to national and international reports on racist violence in Bulgaria, which revealed that the Bulgarian authorities generally did not investigate the racist nature of those cases.

It seems that it may be possible to prove that a protected group is dispropor- tionately affected even where no statistical data is available, but the available sources are reliable and support this analysis.

Example: The case of Opuz v. Turkey involved an individual with a history of domestic violence who had brutalised his wife and her mother on several occasions, eventually murdering the mother.[697] The ECtHR found that the state had failed to protect the applicant and her mother from inhuman and degrading treatment, as well as the latter’s life. It also found that the state had discriminated against the applicants because the failure to offer adequate protection was based on the fact that they were women. It came to this conclusion in part based on evidence that victims of domestic violence were predominantly women, and figures illustrating the relatively limited use the national courts had made of powers to grant orders designed to protect victims of violence in the home. Interestingly in this case, there were no statistics presented to the ECtHR showing that victims of domestic violence were predominantly women, and indeed it was noted that Amnesty International stated that there were no reliable data to this effect. Rather, the ECtHR was prepared to accept the assessment of Amnesty International, a reputable national NGO and the UN’s Committee on the Elimination of Discrimination Against Women that violence against women was a significant problem in Turkey.

Note that statistical data may not always be necessary to prove cases of indirect discrimination. Whether statistics are necessary to prove a claim will depend on the facts of the case. In particular, proof as to the practices or beliefs of others belonging to the same protected category may be enough.

Example: In Oršuš and Others v. Croatia,[698] certain schools had established classes which dealt with reduced curricula as compared to normal classes. It was alleged that these classes contained a disproportionately high number of Roma students and therefore amounted to indirect discrimination on the basis of ethnicity. The government contended that these classes were constituted on the basis of competence in Croatian, and that once a student reached adequate language proficiency, they were transferred to the mainstream classes. The ECtHR found that unlike the D.H. case, the statistics alone did not give rise to a presumption of discrimination. In one school 44 % of pupils were Roma and 73 % attended a Roma-only class. In another school 10 % were Roma and 36 % of them attended a Roma only class. This confirmed that there was no general policy to automatically place Roma in separate classes. However, the ECtHR went on to state that it was possible to establish a claim of indirect discrimination without relying on statistical data. Here, the fact that the measure of placing children in separate classes on the basis of their insufficient command of Croatian was only applied to Roma students. Accordingly, this gave rise to a presumption of differential treatment.

It is also important to note that data and statistics can only be compared when they are available. In this context, under EU law, the Commission published a recommendation[699] in March 2014, focusing on pay transparency. The recommendation aims to propose measures for the Member States to facilitate wage transparency in companies, such as improving the conditions for employees to obtain information on pay or the establishment of pay reporting and gender neutral job classification systems from companies, among others.

Also according to the ECSR, States Parties must promote positive measures to narrow the pay gap, including measures to improve the quality and coverage of wage statistics.[700]


681. ECtHR, Hoogendijk v. the Netherlands (dec.), No. 58641/00, 6 January 2005.

682. CJEU, C-171/88, Ingrid Rinner-Kühn v. FWW Spezial-Gebäudereinigung GmbH & Co. KG, 13 July 1989.

683. CJEU, C-184/89, Helga Nimz v. Freie und Hansestadt Hamburg, 7 February 1991.

684. CJEU, C-33/89, Maria Kowalska v. Freie und Hansestadt Hamburg, 27 June 1990.

685. CJEU, C-343/92, M. A. De Weerd, née Roks, and Others v. Bestuur van de Bedrijfsvereniging voor de Gezondheid, Geestelijke en Maatschappelijke Belangen and Others, 24 February 1994.

686. Opinion of Advocate General Léger of 31 May 1995, paras. 57-58 in CJEU, C-317/93, Inge Nolte v. Landesversicherungsanstalt Hannover, 14 December 1995.

687. CJEU, C-127/92, Dr. Pamela Mary Enderby v. Frenchay Health Authority and Secretary of State for Health, 27 October 1993.

688. Denmark, Supreme Court, Case 28/2015, 14 December 2015, see the English summary in: European Equality Law Review (2016), vol. 1, p. 84.

689. CJEU, Joined cases C-4/02 and C-5/02, Hilde Schönheit v. Stadt Frankfurt am Main and Silvia Becker v. Land Hessen, 23 October 2003.

690. CJEU, C-1/95, Hellen Gerster v. Freistaat Bayern, 2 October 1997.

691. CJEU, C-527/13, Lourdes Cachaldora Fernández v. Instituto Nacional de la Seguridad Social (INSS) and Tesorería General de la Seguridad Social (TGSS) [GC], 14 April 2015.

692. CJEU, C-167/97, Regina v. Secretary of State for Employment, ex parte Seymour-Smith and Perez, 9 February 1999.

693. ECtHR, Di Trizio v. Switzerland, No. 7186/09, 2 February 2016.

694. ECtHR, D.H. and Others v. the Czech Republic [GC], No. 57325/00, 13 November 2007.

695. Ibid. paras. 18 and 196-201.

696. ECtHR, Abdu v. Bulgaria, No. 26827/08, 11 March 2014.

697. ECtHR, Opuz v. Turkey, No. 33401/02, 9 June 2009.

698. ECtHR, Oršuš and Others v. Croatia [GC], No. 15766/03, 16 March 2010, paras. 152-153.

699. European Commission Recommendation 2014/124/EU of 7 March 2014 on strengthening the principle of equal pay between men and women through transparency, OJ L 69, 8.3.2014.

700. ECSR, Conclusions XVII-2 (2005), Czech Republic.


6. Procedural issues in non-discrimination law

6.1. Shifting the burden of proof

6.2. Circumstances irrelevant for the finding of discrimination

6.3. Role of statistics and other data

6.4. Enforcement of non-discrimination law

Leave a Reply

Your email address will not be published. Required fields are marked *