End of the line for the 2011 Census long form

With the announcement that the Fédération des communautés francophones et acadiennes will not appeal the Federal Court's ruling concerning the 2011 Census long form, it's all over now for the 2011 Census.

For the first time since Confederation, that's 143 years ago, the decennial census will not include questions on housing, religion, education, race and occupation of each person (1871 Census), birthplace, citizenship and period of immigration (1901 Census) and all the other population characteristics that make the census such a powerful source of information, beyond basic demographics. Instead, the 2011 Census will be a very basic affair, just 8 questions. For this, Canadian taxpayers will pay $550 million dollars over the full eight year cycle it takes to plan, develop, run and publish the results of a census of population. The full census would have cost $80 million more, for a total of $630 million. That's right, over 85% of the cost of taking the census is taken up just finding and counting the population. It's very cost effective to ask the extra, mandatory, questions to 1 in 5 households as part of the census.

Now instead, that $80 million, plus a promised additional $30 million, will be used to conduct the untested, voluntary National Household Survey, which will be sent to one in 3 households. Census systems, which were completely re-engineered for the 2006 Census to enable a mail-out/mail back method of collection, with an Internet response option, will need to be re-jigged to trigger the mailout of the NHS questionnaire upon receipt of a completed Census questionnaire. This wasn't the methodology that was used in the National Census Test in 2009, which was supposed to be a dress rehearsal for the real thing.

So this is definitely a high risk gambit, which may not only produce poor results for the $110 million spent on the NHS, but which could jeopardize the $550 million being spent on the Census itself. The previous, previous Chief Statistician, Ivan Fellegi, used to say that there is only one way of running a census: running scared. It's a huge beast, which can easily spin out of control, with a high potential for huge cost overruns and unacceptable undercounts. That's why everything is tested and tried in advance, extensive consultation is conducted and high profile endorsements are sought.

I don't know about you, but after filling out and returning my mandatory Census questionnaire, I am not sure I would be that enthusiastic about receiving another, longer questionnaire. Having performed my civic duty, I might be tempted to give it a pass, especially as it is voluntary. And if I were of the political persuasion that shares the Prime Minister's, and Tony Clement's, and Maxime Bernier's, low opinion of government data collection, expressed clearly and frequently on the national airwaves - it's almost like PSA's for not completing the census - I would very certainly not fill it out. So I think we can predict very low response rates for the NHS and probably more difficulty getting cooperation for the mandatory census itself.

And why are we doing this again? Oh yes,

"We recognize that some people are a bit hesitant regarding their private life," Harper said in the House of Commons on Tuesday in response to a question from Bloc Quebecois Leader Gilles Duceppe. "We intend to work co-operatively with the population. We don't threaten to prosecute them for being hesitant. We work with them as adults."

Except when it comes to Swiss bank accounts, of course.

Prime Minister Stephen Harper vowed Thursday that the government would pursue with "the full extent of the law" Canadians who are using secret Swiss bank accounts to avoid paying taxes.

Anyway, at this time, the fiscally prudent thing to do would be to decouple the NHS from the Census altogether in a bid to reduce the risks for the Census proper. If the content that was previously collected in the census long form is to be collected henceforth on a voluntary basis, do the research and testing that will provide for the best, most cost-effective results, including, in the absence of a mandatory census as a benchmark, how to quantify and correct for the biases that will inevitably be found in voluntary surveys (for example, it's pretty safe to predict that Conservative supporters will be underrepresented in the NHS as it now stands).

Maybe it's too late. The train has left the station and it's barreling down the track. Let's hope against hope that we are not headed for a train wreck.