Well, 19% is what we found, but that meant also that about 80% said they would. So if you imply that, or take that over, to any....
First of all, let's separate between the census and the sample, because they're two different things. Nobody, I think, is talking about changing the census. The sample, which, by the way, 80% of Canadian households don't respond to, is the thing that we're talking about.
You could imply from that research, and it's just very preliminary, that 20% of that 20% probably wouldn't fill out the form. But in my experience, a survey response rate of 80% is better than anything I've ever seen. I agree with the other panellists, though, that there's a lot of work that has to be done in terms of calibrating all of this and sorting through what the potential biases are. But it's not impossible. These are problems that statisticians deal with every day.