J.I. v. Croatia: Violence against Roma women – discrimination not an issue?

Written by

J.I. v. Croatia (8.09.2022) is the first European Court of Human Rights case arising from domestic violence against a Roma woman and the second related to sexual violence against a Roma girl. Police were neglectful. The intervener exposed systemic police normalisation, and disregard of, Roma intra-community violence against women. 

A procedural violation of Article 3 ECHR was found: repeat victimisation through death threats by a rapist father was not investigated. Recognising J.I.’s ‘particular vulnerability as a Roma woman and a victim of serious sexual offences’, the Court acknowledged that leaving her to fear further assaults was inhuman treatment.

Yet, the Court undermined its intersectional vulnerability recognition by disclaiming discrimination as an issue. It declined to examine whether the victim was neglected based on ethnicity. In a summary fashion characteristic of its covert discrimination jurisprudence, the Court disallowed any issue under Article 14 ECHR without sufficient reasoning. While the judgment is commendable in important respects, it is nevertheless symptomatic of a general fault – discretionary gate-keeping of Article 14, marginalising equality issues.


J.I. survived multiple rapes by her father. He was imprisoned. Extremely traumatised, suicidal and self-harming, J.I. changed her name, appearance, and domicile.

While on leave, her abuser threatened to kill her, through relatives. J.I. repeatedly asked for police protection. None was provided; no investigation ensued. Officers were dismissive. J.I. reported police misconduct. They persisted not investigating the threats, breaching domestic law.

Third party

ERRC submitted that Roma girls/ women are disproportionately targeted for violence against women, while police and other institutions normalise their abuse as inherent to their communities, failing to protect them. Evidence of such police bias is rare. “[B]ecause stereotypes meant that gender-based violence against Roma girls and women was largely ignored, comprehensive data about police failures to protect them would be even harder to produce’. A hypothetical comparator is needed: police withhold the response they would give non-Roma victims. (§74-9)


The Court acknowledged J.I. as ‘a highly traumatised young woman of Roma origin […] the victim of appalling sexual abuse by a close family member at a very early age’. (§86) The death threats were psychological violence. Despite not being direct, they caused a vulnerable victim intense fear. (§88) Considering her excessive trauma and feelings of powerlessness facing repeat victimisation, her treatment was inhuman under Article 3. (§89)

A form of domestic violence, the threats triggered investigation duties. (§91) Instead of discharging those, existing also under domestic law, police maintained they could do nothing unless the abuser confronted J.I. (§92) No comprehensive analysis, required in domestic violence cases, was attempted. (§99) No prosecutor was informed, no investigation opened, despite J.I. reporting threats thrice. (§95-6, 100)

The Court did not rule on the failure to protect J.I. from repeat victimisation: there was ‘no need’ to. (§101)

Regarding discrimination, ‘neither the circumstances as submitted, nor any relevant evidence such as statistical data, substantiate the allegation’. (§97) Despite J.I.’s ‘particular vulnerability on account of her sex, ethnic origin and past traumas’ and the police’s ‘failure to perceive the seriousness of [her] allegations’ being ‘in blatant disregard’ of domestic law, there was ‘no separate issue under Article 14’. (§108)

EUR 12,000 was awarded for non-pecuniary damages.


Judge Wojtyczek contested the violation: the authorities’ assessment of J.I.’s allegations as ‘[in]sufficiently substantiated’ was possibly correct.

Judge Derenčinović considered a ruling on the protection failure necessary. Duties to prevent domestic violence applied particularly to repeat victimisation. ‘[N]ot a single measure of victim protection was taken’, severely impacting J.I. and possibly encouraging domestic violence. The authorities knew J.I. faced a real and immediate risk, and did not manage it as required: no risk assessment, no proactivity, no inter-agency cooperation. Personnel were dismissive, ignoring the perpetrator’s serious-offender profile and the victim’s vulnerability.


The judgment is respectable – victim-responsive, gender-sensitive to repeat victimisation, recognising the threats amounted to violence despite being mediated. Few other judgments qualify DV as inhuman treatment: Volodina v. Russia (§75), Eremia v. Moldova (§54), Mudric v. Moldova (§45). In all those, victims were caused to fear repeat victimisation. The Court is mindful of such intimidation’s impact on DV victims. There may be no other violence against women judgments (outside detention contexts) specifying treatment was inhuman. J.I. is victim-centered, validating J.I.’s experience, articulating the distress and defencelessness inflicted. Significant compensation is awarded.

The Court reiterates its strong stance on DV, stipulating authorities must comprehensively investigate such threats, heeding the history of abuse. It acknowledges J.I.’s heightened vulnerability as intersectional, ethnicity as central as gender and severe early-age victimisation.

Yet, it limits vulnerability implications to investigation duties, withholding an Article 3 substantive breach discussion. J.I.’s exposure to non-protection warranted that, as Judge Derenčinović argued. The Court should have scrutinised the authorities’ failure to: consider whether to grant the perpetrator leave in the same (part of) town where J.I. lived; inform her of his leave; caution him to abstain from (indirectly) contacting her; instruct police to be responsive should J.I. report concerns; appoint gender-sensitivity trained officers as her contact; warn the perpetrator after his first threats; inform J.I. of that and of her victim rights.

Importantly, instead of being perceptive of J.I.’s vulnerable identities as a likely inequality factor, the Court dispatches her discrimination allegations. It discounts any Article 14 issue merely because the alleged inequality is covert, absent a racist utterance (‘neither the circumstances as submitted […] substantiate the allegation’) and statistics on police neglect of such victims.

Yet, covert discrimination is as real as, and more common than, overt discrimination. Open bias should not be required to apply Article 14. Statistics are not required to find discrimination (ex., Oršuš and Others v. Croatia, §75). The intervener documented a context of police neglect of victims like J.I. and a dearth of proof. Such findings by an established organisation amount to qualitative equality data. Such data, together with the case fitting the inequality pattern, the victim’s vulnerability, the police omission’s blatant unlawfulness (Court-recognised), and the lacking explanation for it suffice for Article 14 application, if not for a prima facie case.

In this context, the hypothetical comparator the intervener suggested would strengthen an inference of discrimination: an average (non-Roma) victim would not be treated as poorly. A ‘would-be-treated’ comparator is an established direct discrimination instrument (ex., Article 2.2.a, Directive 2000/43). (It has been integral to UK equality legislation since the 1970s.) The Court has never referenced it.

A hypothetical comparator does not require evidence pertaining to a particular individual but instead relies on legitimate generalisations/ assumptions, such as, for example, that police would not normally withhold the response to criminal threats they are legally required to have. 

It is rational to expect the State to refute thusly inferred discrimination by showing a race-neutral reason for the failure, rather than expecting the victim to prove bias:

“[B]ecause people rarely advertise their prejudices and may not even be aware of them, discrimination has normally to be proved by inference rather than direct evidence. Once treatment less favourable than that of a comparable person […] is shown, the court will look to the alleged discriminator for an explanation. The explanation must […] be unrelated to the race or sex […]. If there is no, or no satisfactory explanation, it is legitimate to infer that the less favourable treatment was on racial grounds.” (House of Lords – Regina v. Immigration Officer at Prague Airport, 2004)

In discrimination law, it is standard to use inferences as affirmative proof rather than expect direct evidence:

“Direct evidence of a decision to discriminate on racial grounds will seldom be forthcoming. Usually the grounds of the decision will have to be deduced, or inferred, from the surrounding circumstances. […] Members of racial groups need protection from conduct driven by unrecognised prejudice as much as from conscious and deliberate discrimination.” (House of Lords – Swiggs and Others v. Nagarajan, 1999)

In leading jurisdictions, it has been settled jurisprudence for decades that different treatment (including compared to how another would be treated) and racial difference indicate possible racial discrimination. In such circumstances, adjudicators expect the respondent to convincingly explain. Unless they do, discrimination is legitimately inferred, as a matter of “almost common sense”. (May L.J. in North West Thames Regional Health Authority v Noone: CA 1988) Such a conclusion should be reached “on the balance of probabilities, bearing in mind the difficulties which face a person who complains of unlawful discrimination.” (Neill L.J. in King v. Great Britain-China Centre [1991] I.R.L.R. 518, cited by House of Lords – Strathclyde Regional Council v. Zafar, 1997)

(While some of these judgments pertain to employment, the principles are equally valid for service provision and many other areas under EU law. There is no reason not to apply them to all non-criminal cases.)  

The Court has yet to adopt these standards. It routinely recalls it is hard for States to prove racial motives in crime (ex., M. and Others v. Italy and Bulgaria, §176), but has not recognised it is harder for victims, for non-criminal acts too. In J.I., the Court does not mention the shifting burden of proof, which should invariably apply in discrimination cases. It ignores the intervener’s hypothetical-comparator suggestion and the systemic issue – authorities tending to disregard violence against women amongst Roma.

The Court overlooks its jurisprudence as a source on this general issue. In M. and Others (cited), the respondents defended their omission to investigate alleged rapes of a Roma girl within a Roma family, referencing a ‘specific’ ‘context of a Roma marriage’ involving payment for the bride in exchange for ‘consummation’. (§§92-3, 123, 135, 137) The Court explicitly discarded this ‘Roma custom’ justification, finding an Article 3 procedural violation. (§105-7) In J.I., it drew no conclusions from that illustration of official tendencies to normalise rather than investigate intracommunity Roma girls’ abuse attributed to Roma culture.  

Contemporary discrimination law expects to lack direct evidence, validating a claimant’s inability to adduce it rather than using that against her. Inferences, shifting the burden of proof, hypothetical comparators and other tools apply by default to offset this objective evidentiary shortage. The use of those tools should be standard practice under Article 14. The absence of direct proof should not prevent Article 14 application.   

In M. and Others (cited), the Court took an analogical approach, mutatis mutandis, to the sexual violence complaint, addressing it despite the lack of medical evidence: it was logical that the applicant should not have such evidence as she was allegedly held captive. (§101) Therefore, such evidence could not be expected from her. Her testimony and the seriousness of her allegations before the authorities sufficed for a reasonable suspicion that she was ill-treated as alleged. (id.) Article 3 was applied.

The seriousness of J.I.’s ill-treatment – inhuman – and the ensuing seriousness of her allegation that it was identity-based should have similarly resulted in applying Article 14.        

Indeed, lacking proof of discrimination should be held against respondents where (partly) due to their not investigating. In M. and Others, the Court noted ‘the lack of investigation by the Italian authorities has led to little evidence being available to determine the case’. (§152) In Volodina (cited), it faulted Russia for Volodina’s inability to present data showing discrimination – the authorities failed to collect such data. (§118) In J.I., had the authorities addressed her police misconduct complaint, there might have been information to support or refute an inference of bias against her. The Court omitted that.

J.I. is no departure from the Court’s discrimination jurisprudence. Deciding it differently – addressing inequality – would require the jurisprudence to expand, incorporating the above standards. 


J.I. is a creditable judgment, demonstrating violence against women victimhood awareness.

Less strong on victim protection standards, it also reflects deficits in handling discrimination allegations. Scholars recognise ‘political uses of the international human rights evidence system, including dismissals of politically sensitive complaints on the pretext they are not sufficiently evidenced by the victim.’

Roma women’s structural inequality is not a non-issue. Their discrimination allegations should be contextually analysed, using established instruments to compensate for bias obfuscation, not blaming victims for it, yielding well-reasoned responses they are entitled to.

Print Friendly, PDF & Email

Leave a Comment

Comments for this post are closed