Tony Clement's office knew Statistics Canada felt a voluntary census would fall short of the mark even as the Industry Minister defended scrapping the mandatory long form by suggesting the agency had embraced the change, newly released documents show.
In a March of 2010 e-mail to Mr. Clement's office, a senior Statscan official advised that a self-administered voluntary survey would yield an initial response rate of only 50 per cent.
She added that with follow-up work and sufficient resources for interviewers, the agency could push that feedback rate to between 65 to 70 per cent, "which is still not an acceptable outcome for a census." Mandatory census response rates are much closer to 100 per cent.
Despite this caution, Mr. Clement as recently as July was invoking Statscan to defend the Conservatives' controversial decision to scrap the mandatory long-form census in the name of protecting privacy. For decades, more than one-fifth of households have been required to answer its more than 50 detailed questions about work, home life, ethnicity and religion.
The Conservatives this week released more than 300 pages of confidential records on the census controversy to the Commons industry committee, painting a picture of Ottawa's frantic efforts to sell the move to Canadians despite mounting opposition.
One transcript of Mr. Clement's remarks at McGill University in mid-July details how he left the impression Statistics Canada was confident the voluntary survey could substitute.
"We've come up with a way that is statistically valid, that Statscan feels can work," he told a Montreal audience on July 15. He said the agency gave him options to replace the mandatory long-form after he requested ones that "will not degrade the data."
Addressing his critics, Mr. Clement declared: "I would say to those who are the self-proclaimed experts on this, if you trust Statistics Canada why don't you trust the option that they put forward to obtain the data that businesses and municipalities deem to be necessary."
E-mails and memos also made public show Canada's former chief statistician, Munir Sheikh, intended to publicly acknowledge shortfalls in the substitute voluntary survey on the day he ultimately resigned from the job.
This replacement survey is being sent to one-third of households but there are no threats of penalties or jail times for ignoring it.
Documents reveal Mr. Sheikh had originally intended to tell an employee town hall at Statistics Canada on July 21 that the Harper government's voluntary alternative was in some ways inferior to the mandatory long-form census. He never delivered this speech because after cancelling the town hall, the veteran bureaucrat quit that day.
"The National Household Survey, by its very design, will not be the equivalent of a mandatory long-form census," a draft version of his speech says, adding that a "census, by its design, is able to meet a broader range of data demands than a survey."
Mr. Sheikh intended to say that some of the consumers of Statscan data would be disappointed in what the voluntary survey yielded.
"The [survey]will meet the needs of many users but will not provide useful data to meet the needs of other users of the mandatory long-form census data."
Mr. Sheikh declined a request for comment Tuesday. The veteran bureaucrat resigned after being frustrated at Mr. Clement's repeated claims that Statscan had assured him a voluntary survey could adequately replace the mandatory long form. Upon resigning, he issued a statement saying it was not a substitute.
Documents released show the Privy Council Office as well as Mr. Clement's office were actively suggesting to Mr. Sheikh how he should word a public statement on the census changes - another document that was never released because he resigned.
"How about this?" asked one senior Privy Council staffer in an e-mail suggesting changes to Mr. Sheikh's planned statement. "A little softer and ends on a positive note while saying the same thing."
Mr. Clement's office said they never at any time tried to control what Mr. Sheikh said regarding the census changes.
Mr. Sheikh, in an interview with CBC TV's Power and Politics Tuesday, said the reliability of much of Statistics Canada's work will suffer without a mandatory long-form census. "The census is a benchmark for a lot of the surveys we do at Statistics Canada. If we don't have that benchmark, we really have no way of knowing whether or not the data produced even by those surveys is something we can trust."