Because of the absence of statistics known as “ethnic” in France, social scientists do not have at their disposal data on possible discriminations produced by criminal law judges. To compensate for this dearth, Fabien Jobard, researcher at the French National Centre of Scientific Research (CNRS), has examined decisions handed down by a court in the Paris area with respect to acts of insulting conduct, rebellion and violence against agents of the public authority. His study helps to place in much clearer perspective the commonplace by which judges are indifferent to skin colour in their judgments.
Little is known in France about possible acts of discrimination produced by judicial decisions. In Anglo-Saxon countries, “race and sentencing” studies are superabundant. In these countries, researchers have at their disposal pre-formatted statistical data by the records of ethnic characteristics the persons judged or the victims declare voluntarily. The ability to access ethnic statistics simplifies the work of objectifying discriminatory acts. This is not possible in France, where the task consisting of determining whether there is discrimination in judgments rendered requires the researcher to generate the statistics himself.
It is to this task that Sophie Névanen and I have addressed ourselves. [1] We have assembled the judgments delivered by the District Court of Melun (Tribunal de grande instance de Melun) from 1965 to 2005, concerning ourselves with a sole type of offence: offences against persons invested with the public authority (referred to as IPDAP), whether they are acts of insulting conduct, rebellion, insult and rebellion, violence (and combinations of these offences). The data on IPDAP cases tried, without including the offences, make it possible to develop a clearly defined set of qualifications, which is relatively unpolluted by legislative evolutions, as can be the cases for offences of violence or robbery, complicated by diverse aggravating circumstances over the succession of laws that compromise their historical legibility.
We have thus assembled a set of 864 defendants tried in a criminal court in an outlying Parisian suburb, which we characterized according to their origin and the consonance of their names, forming three groups (modelled on the classification used by the Criminal Investigation Department): the “European” group, the “North African” group and the “Black” group. A defendant born in France, bearing a typically African family name or first name, belongs to the African group; a defendant born in the Maghreb belongs to the “North African” group (except if he bears a French Catholic name, in which case he is considered to be a pied-noir - a French national born in colonial North Africa and therefore belonging to the “European group”); a defendant born in France and bearing an Arab or Berber name belongs to the “North African” group, etc. This classification is rudimentary because it neither allows the characterization of Black defendants with an Arab name, nor of the Black defendant born in continental France and bearing a typically French Catholic family and first name (such as the football player Thierry Henry).
Notwithstanding all this, during the period under consideration, 62% of defendants tried for a separate IPDAP offence refer, over the period under consideration, to the “European” group, 20% to the “North African” group, 15% to the “Black” group (and 2.2% to other groups). First discovery: these groups are undeniably unequal in regard to sentencing. Firm prison sentences were delivered against 13.6% in the “European” group, but against 23.6% in the “North African” and 25.4% in the “Black” groups. To put it differently, the risk of firm imprisonment is nearly twice as high for defendants from the “North African” and “Black” groups as for those from the “European” group. However, these groups may be different from one another, and composed differently: it is necessary to examine the “composition effects” or the “structural effects” (please see statistical box below), because if we end our analysis with groups constructed in this manner, making a single sorting, we risk concealing a great deal and saying too little.
Glossary
*Structural effect : When a population is divided into subpopulations, it may occur that a characteristic evolves in one direction in each subpopulation, while evolving in the opposite direction for the whole population. This paradox can be explained because the sizes of certain subpopulations increase while others diminish: that is the structural effect.
For example, the salary of each profession may stagnate (or increase slightly) while the average salary increases strongly; this occurs if the highly qualified professions (the best paid) become more and more numerous, and vice versa, unskilled employment (the least paid) becomes more rare. A contrario, the variation, holding structure constant, is calculated as a weighted average of average variations, the weights being the size of each subpopulation. (Source: INSEE, Nomenclature).
* Multivariate Analysis: statistical analysis of relations that may exist between several (response) variables.
Thus, it is necessary to first distinguish the separate offences that belong to the class of offences against persons invested with the public authority (IPDAP). Sentences stipulated for violence are greater that those for rebellion, etc. In our court, the risk of firm imprisonment that was observed is 1 out of 10 for simple insulting conduct, 1 out of 5 for rebellion or for insulting conduct + rebellion, and 1 out 3 for violence. Now, the accused individuals in the “European” group are more frequently arraigned for offensive conduct and less frequently for violence (and inversely for the other groups). Also, the defendants in the “European” group are more frequently present at the hearing than other groups; and this type of judgment, known as “contradictory”, is always more favourable to the defendants. Finally, it is among the groups of “Blacks” and “North Africans” that are found the highest proportion of defendants being judged as repeat offenders (most often brought to court for immediate trial): an accused individual who is judged in immediate trial incurs nearly one chance out of two of being sentenced to firm imprisonment.
The impact of these structural and compositional effects increases when one combines all these variables to determine the sentence handed down, all else held constant, by means of a multiple regression, and it becomes clear that the “group” variable does not have, in itself, any influence. In reality, this variable contains the other variables; it is only a reflection of the other variables, which in fact are the ones that determine the decision of the judge. Thus, in spite of appearances that are nevertheless spectacular, judges are indifferent to the colour of the skin or to the origin of the defendants.
On the other hand, the situation is different in regards to policemen. Since they are victims, they have the possibility of becoming plaintiffs in the proceedings and to sue for damages in compensation for the moral harm suffered. Now, 37% of accused individuals in the “European” group face police officers who exercise their right to appear in court as a plaintiff, as opposed to 51% of defendants in the “North African” group and 46% in the group of “Blacks”. And in this case, the multivariate analysis shows, all else remaining equal, that the “group” variable does in fact have an influence of its own (second statistical box below). While the decisions of the judges seem to be impervious to the physical appearance of defendants, personal decisions of the policemen - subject to further study over a larger sample - are not.
What should be the conclusion? First, that it is necessary to be prudent with statistics. A scientific answer to a political question (“Is there discrimination?”) is fairly complicated: it does not suffice to discern a statistical discrepancy, however large it may be: it is necessary to study its composition and its structure. Moreover, the statistics record persons who have already been selected – by the public prosecutor’s office, by the criminal investigation officer and by the policeman who decided the arrest. Consequently, our study does not make it possible to say that “Blacks” or “North Africans” commit the worst offences against policemen more often than others. It is necessary to say that “North Africans” and “Blacks” are more frequently judged for more serious charges. Finally, this study provides us with a major lesson on the more general mechanism of judicial discrimination. Judges are blind to differences. In the process, nevertheless they aggravate them. This paradox is the result of their position as the last link in a penal chain that has produced their decisions upstream.
Judges know that they can decently reserve imprisonment solely for their clientele: those who have already been in prison. A defendant who has already been sentenced runs a great risk of being sentenced again. Furthermore, a “socially integrated” defendant (having a permanent work contract, a stable family situation, etc.) will received clemency from the judge, who knows that prison would condemn the accused as much on the social as on the criminal plane. (This realism in the profession places the failure of the penitentiary’s mission on display every day.) This is what our study thrusts into the spotlight: “Blacks” and “North Africans” make up the greater part of the criminal clientele, that such a particular population, the main part of whose socialization is, very rapidly, assumed by the judicial system.
Fabien Jobard, « Do Criminal Judges Discriminate? »,
Books and Ideas
, 24 July 2008.
ISSN : 2105-3030.
URL : https://laviedesidees.fr/Do-Criminal-Judges-Discriminate
Nota Bene:
If you want to discuss this essay further, you can send a proposal to the editorial team (redaction at laviedesidees.fr). We will get back to you as soon as possible.
[1] Fabien Jobard and Sophie Névanen, “La couleur du jugement. Discriminations dans les décisions judiciaires en matière d’infractions à agents de la force publique (1965 – 2005)”, Revue Française de Sociologie, Vol. 48, 2007/2, pp. 243 – 272.