The first World Statistics Day was celebrated at Statscan on Wednesday with guest speakers, free coffee and enough vanilla cake to feed 400 of the agency’s 6,000 employees. Statistically speaking, that means 93 per cent of staff had to go without.
Not eating cake is just the latest disappointment in a tumultuous year at Canada’s national statistics agency.
The mandatory long-form census was cancelled in June, chief statistician Munir Sheikh resigned in July and now new documents made public by the agency say the voluntary survey that will replace the long form next year simply won’t measure up.
Statistics Day’s dark mood was captured in a mock funeral for the mandatory long-form census, which the Statistical Society of Canada posted on YouTube. The document was interred as a voiceover expressed sorrow at “her sudden and untimely passing.”
Don McLeish, president of the society, said the irony of the United Nations choosing to honour the contributions of official statistics in the year that Canada turned its back on one of its best tools was too rich to ignore.
On the World Statistics Day website, which has a news section populated by stories marking celebrations and declarations of the day around the world, Canada is distinguished by having the only contentious headline: “Canada’s Chief Statistician Resigns on a Matter of Integrity.”
In a document posted to its website this week, Statistics Canada offers implicit criticism of the decision to scrap the mandatory long form. It says its replacement, the National Household Survey, will produce “useful and usable data” but will not be as good as its predecessor.
The document says sampling bias in the voluntary survey could skew the results to either overestimate or underestimate certain populations. No other country has ever attempted a survey on the scale of the National Household Survey, and the new methodology has been introduced “relatively rapidly and with limited testing,” it says. Only half of recipients are expected to complete the survey.
“It means society will look a whole lot better. It’ll be richer, better-educated and more conservative, I suspect, when the results come in,” Prof. McLeish said.
One of its simulations, which did not use mitigation strategies to compensate for sampling bias, found that Toronto would look, on paper, like a very different city. Blacks would be underrepresented by 13.2 per cent and Chinese overrepresented by 17.6 per cent. Registered Indians would be underrepresented by 11.7 per cent, and construction workers would be underrepresented by 9.4 per cent. The same is broadly true in the other test cities of Winnipeg and Bathurst, N.B.
In a system dependent on enthusiasm for filling out forms, it’s perhaps no surprise that bureaucrats would loom larger. In all three cities, public administrators would be overrepresented. The data suggest that poorer people, immigrants, aboriginals and visible minorities would all be undercounted unless new strategies can be created to overcome the selection bias.
Even if those issues are addressed, a voluntary survey will result in a significant breakdown of statistical continuity, experts say.
Gabrielle Beaudoin, director of client services at Statscan, said employee morale at the agency is high despite the controversy over the census. Employees received a note of encouragement Wednesday from Industry Minister Tony Clement. And although there wasn’t enough cake for everyone, there was a lot of free coffee and tea, Ms. Beaudoin said.
“The idea was not to make this an event that would cost a lot,” she said. “The idea is just to underline that it’s a special day.”
