At Statistics Canada, as I mentioned to Mr. Godin, we conduct an enormous number of surveys on various topics. When we conduct surveys, often funded by various federal departments, we are able to ask the same questions in all provinces and of all respondents. To minimize the volume of responses, of course, we often try to use administrative files. If we already have information in administrative files, why would we ask questions in the context of surveys?
The problem is better documenting the language in which people study. For example, we know how many children attend immersion schools, how many children attend minority schools, but we don't know, based on those files, the mother tongue of those people. That information is not available. We had to wait for the Survey on the Vitality of Official Language Minorities in 2006 to really establish that 50% of children who have a French-speaking parent attend a minority school. Otherwise, we didn't know that because, in certain cases, anglophone children were attending minority schools because there was no immersion program. These issues are quite—