I repeat, the response rate that is being quoted by Statistics Canada as 98% is essentially the questionnaires we got back within the universe of the questionnaires we sent out. There are differences, as I said, between the long form and the short form and within the different geographies.
Even within the questionnaires themselves, of course, there is item non-response, and that comes out as we put out all the different statistics on labour mobility or immigration or language or whatever. What happens is that for certain cells you can have fewer responses than you do for others. When those data are released, the quality identifiers for each of those.... There's also sampling, as you know, for the long form; not every Canadian household fills that out. All the responses, as well as that sampling, are taken into account, and a quality identifier is released with the information.