Thanks, Jim.
Let me start by asking, what is distinct about the mandatory long-form census?
First, the response rate is very high, so the measure is free from bias to the maximum extent possible. This is just as important as the size of the sample.
Second, the questionnaire includes a great variety of questions. These are important not only because they're important individually, but because of the way they're combined. People have talked about cutting down the length of the survey, but that cuts down the ability to look at a variety of different things. For example, you might be interested in the poverty rates of children, but as a global statistic, that's a lot less interesting, in a way, than knowing how poverty rates of children vary across the country, among ethno-racial groups, for aboriginal communities, and so on.
Third, the long-form sample is very large, so we get accurate statistics from small communities in rural areas for individual racialized groups, and so on. None of Statistics Canada's smaller surveys is a substitute for the census for this reason.
Fourth, because there is a great deal of overlap in the questions from one census to the next, it is possible to measure change over time very accurately. Often the change is as important an indicator as the absolute level. So we talk about the aging of the workforce, not the age of the workforce. The critical thing is the change.
Much of the information obtained from the census is not available from any other survey, particularly questions dealing with racialization and immigration.
Finally, questions asked in the census are decided in a painstakingly careful process. The size of the enterprise and its cost are so large, there's a detailed rationale for the inclusion of every single question.
So the question is, why is there a problem with the proposed voluntary survey? In answer, the critical thing is that the response rate will be between 60% and 75% and the results will be biased by non-response. This is because the people who do not answer the survey are different from those who do. We know the response is lower among young people, more mobile people, poorer people, and so on. The problem is that with the switch from the mandatory census, it's not simply that the content of the survey is changed, but the critical thing is that change from one census to another can't be distinguished from non-response bias when you move from a survey with 4% or 5% non-response to 25% or 30% non-response.
Thank you.