Toby Sanger has a post up at The Progressive Economics Forum discussing the effects of changes the Harper government made to the census. Specifically, the Conservatives eliminated the mandatory census long form and substituted a voluntary National Household Survey despite the warnings of, well, pretty much everyone who knows anything about statistics that the change would damage the reliability of the resulting data.
As Sanger notes, Statistics Canada has declined to publish information it normally would on low income Canadians because the data is unreliable. Instead they've issued a note to let readers know that the info is available on request but should be viewed with much more skepticism than would previously have been required for this kind of data set.
Sanger then points to this post at Economic Justice which notes that even the data StatsCan is releasing is the result of a drastic lowering of the agency's previous standards.
I agree with the first commenter on Sanger's post: this was the real point of the change. Too much of the data coming out of Statistics Canada was providing solid evidence for opponents of the policies promoted by Conservatives and that had to change. Now that evidence is either missing or far less reliable than it used to be.
Let's say a special thanks to Tony Clement who was the relevant cabinet minister at the time and distinguished himself by lying repeatedly during the public debate on the issue.
And might I remind everyone that during that debate, the Fraser Institute was prominent in its support of the government's position and sought to downplay any negative effect it might have. Odd behaviour for a so-called think tank that produces so many reports that, theoretically at least, should depend on accurate statistical information.