Note: following a helpful comment, this post has been edited to correct a mistake on my behalf.
In this blog entry I would like to focus on the Court’s interpretation of the concept ’statistically relevant’. In Oršuš and others v. Croatia, the Court held the following regarding the statistical evidence adduced by the applicants:
“These statistics demonstrate that only in the Macinec Primary School did a majority of Roma pupils attend a Roma-only class, while in the Podturen Primary School the percentage was below 50%. This confirms that it was not a general policy to automatically place Roma pupils in separate classes in both schools at issue. Therefore, the statistics submitted do not suffice to establish that there is prima facie evidence that the effect of a measure or practice was discriminatory.” (§ 152, emphasis added)
The use of statistics to prove the existence of indirect discrimination was introduced in D.H. and others v. The Czech Republic (app. no. 57325/00) and used afterwards in Sampanis and others v. Greece (app. no. 32526/05). This introduction can only be applauded and was a major step forward for the protection of minorities, since it made proving indirect discrimination more feasible.
What is regrettable is the simple conception of statistical evidence the Court uses in Oršuš .
If the percentage is below 50, it does not indicate anything. That is a bit too easy. It demonstrates that the Court does not really grasp the concept of statistical relevance. Why else would such an arbitrary demarcation line as 50% be introduced? My personal opinion is that the Court got stuck in a reasoning of majority rule. After all, nine votes beat eight votes and therefore the majority holds the truth. Neglecting the fact that in different circumstances a basic truth can also be found in a minority is particularly regretful in a case involving vulnerable minorities. Perhaps my point can be further clarified by drawing a – rather hyperbolic, admittedly – comparison to the crime of genocide. Would the introduction of a 50% requirement not strike us as odd in such a context? Of course it would. Because it does not make sense.
Naturally, the main point the Court was trying to bring across in Oršuš was not that statistics only prove something if upon aggregation the relevant data exceeds the 50% demarcation line. The point was that the discrimination needed to be proven beyond the statistical evidence since the latter alone did not offer conclusive proof. But the Court should have made this clearer and should have avoided attempting to be something it is not. A mathematician.
By Stijn Smet