I would just add that Statistics Canada itself has modelled the non-response bias that's likely to occur in three sample municipalities, one of them being Toronto. Very clearly, the modelling that Statistics Canada has done shows that there will be significant bias in particular populations. Low-income families will be under-represented significantly as a result of the non-response bias, as will construction versus public-sector administration jobs. Also, the cultural diversity of the city will not be measured accurately. There will be an over-representation of the Chinese population, and a significant under-representation of the black community and youth. So we know that the bias will be there, and the very populations we hope to serve will be under-represented as a result of the non-response bias.
On November 23rd, 2010. See this statement in context.