France has long decided that this information has no relevance, and has adopted laws that formalise this, and refuse to give relevance to that information. You come with different opinions on that topic, but that's just your opinion. France disagrees that that information should be made available, because that has social and political consequences. It's our democratic choice.


In the long run, we're all dead. John Maynard Keynes

Ok, there is another institute, INED [Institut National d'Études Démographiques, National Institute for Demographic Studies] but it also has no statistics on religious self-identification that I can find.

It does have a journal which, in 1996, published a couple of articles on the topic. (PDF). The cover article, "Describing Minorities" (Décrire les minorités) is written by  demographer Michel Louis Levy, in whose blog I found a recent related entry where he says "France has a cruel lack of surveys on religious behaviours, but it would be catastropic to even consider a census".

Michel Louis Levy: Ethnic  counting and positive discrimination (3 July 2006)

Michel Louis Levy: Comptage ethnique et discrimination positive (3 juillet 2006)

...	...
There are in all this terrifying confusions. We run the risk of seeing the launch of a new and absurd press campaign, analogous to that over ethnic categories, which Le Monde, already, and Le Nouvel Observateur have waged ten years ago, with the help of Hervé Le Bras, against Michèle Tribalat. In my capacity as Communication Director of the INED, I tried to defend her in several papers and interventions, among them the editorial of number 309 of Population & Sociétés, entitled "Describing Minorities". "In the near past", I wrote, "innovative surveys have tackled in good conditions topics with a reputation for being delicate, sexual behaviours and integration of immigrants. These encouraging experiences show that as long as the law, the rules of the craft and  professional ethics are respected, there are no taboo subjects".	Il y a dans tout cela de redoutables confusions. Nous risquons de voir se déclencher une nouvelle et absurde campagne de presse, analogue à celle sur les catégories ethniques, que Le Monde, déjà, et Le Nouvel Obs avaient menée il y a dix ans, avec l'aide d'Hervé Le Bras, contre Michèle Tribalat. En tant que Directeur de la Communication de l'INED, j'avais tenté de défendre celle-ci dans plusieurs papiers ou interventions, dont l'éditorial du numéro 309 de Population & Sociétés, intitulé "Décrire les minorités". "Dans le passé récent", écrivais-je, "des enquêtes novatrices ont abordé dans de bonnes conditions des sujets réputés délicats, les comportements sexuels et l'insertion des immigrés. Ces expériences encourageantes montrent que dès lors que sont scrupuleusement respectés la loi, les règles de l'art et la déontologie professionnelle, il n'y a pas de sujet tabou".
Any survey implies the agreement of the people questioned, and cannot give rise to other than statistical results. France has a cruel lack of surveys on religious behaviours, but it would be catastropic to even consider a census. In the same way, one can make surveys on declared ethnic origins, and even on felt discriminations, as long as these surveys remain specialised. But to classify people according to ethnic origin, within the compulsory survey of the census, can only cause mistrust of public statistics, which have no need for this additional suspicion.	Toute enquête suppose l'accord des personnes interrogées, et ne peut donner lieu qu'à des résultats statistiques. La France manque cruellement d'enquêtes sur les comportements religieux, mais il serait catastrophique de y consacrer la moindre question du recensement. De la même façon, on peut faire des enquêtes sur les origines ethniques déclarées, et même sur les discriminations ressenties, dès lors que ces enquêtes restent spécialisées. Mais classer les gens selon l'origine ethnique, dans l'enquête obligatoire qu'est le recensement, ne peut que susciter la méfiance envers la statistique publique, qui n'a pas besoin de cette suspicion supplémentaire.
...	...

Nothing is 'mere'. — Richard P. Feynman