Electile Dysfunction
Evidence suggests that we all like it long.
Mayors of Canada like it long. It helps them prioritize the next transit or water main development.
Police like it long. It helps them decide how to recruit, to reflect the changing communities they serve.
Medical researchers like it long. It helps them see patterns of disease, which helps find cures.
Business entrepreneurs like it long. It helps them launch a new product or service.
School boards like it long. It helps them pinpoint where to build new schools.
Boy Scouts and other after-school programs like it long. It helps them target the neighbourhoods that need more options for youngsters.
Large foundations and charities like it long. It helps them prioritize where to place their money.
Clergy like it long. It helps them discern where their flocks are headed.
Public health officials like it long. It helps them roll out immunization strategies during epidemics.
The international community likes it long. It helps us all look over each others’ shoulders and compare notes in reliable, consistent ways.
Even Stephen Harper secretly likes it long. It helps his team target which “ethnic†or otherwise vote-rich ridings to rustle up for the election.
Over the years we’ve learned that evidence from the mandatory long-form census questionnaire helps us make sharper, more effective decisions.
Then along comes Stephen Harper and decides to cut those practices short, for Canadians from every walk of life and governments of every political stripe.
It seems a classic case of denial. After all our PM uses the data as much as the rest of us, perhaps more. And, like a moth to the flame, he’s running headlong into the story one more time.
Stephen Harper had his pick of what day to send Canadians to the polls. He chose May 2nd. Irony of ironies, May 2nd is also the day the Census starts getting mailed out to you and me.
Well, that’s the Census short-form questionnaire, the one everyone has to answer. It’s basically a head-count of where the males and females of our nation live, how many people live with us, our ages, and whether we speak English or French.
The long-form questionnaire will go out about a month later. It’s the one the Harper Team says is intrusive and that you shouldn’t have to answer. Never mind they are asking all the same questions in the new, improved model.
At an additional cost of $30 million, the government will send the questionnaire to one in three households instead of one in five. (That means the odds of your household receiving it just jumped from twice, maybe three times in your life to 4 or 5 times.) The difference is that instead of numerous follow-ups with households that don’t reply, there will be just one follow-up….and an ad campaign. Instead of it being mandatory, an act of citizenship that helps serve your interests and mine, it will be “whateverâ€.
We’ve just been through the worst global economic crisis in 80 years. Almost half a million full-time jobs were wiped out in the space of 9 months.
So, almost three years into it, how did Canadians weather the storm? How many of us are back on track, compared to 2006? How many of us aren’t?
The Harper government doesn’t think we need to know in any reliable way.
The government has been fully briefed by Statistics Canada about the implications of switching the census long-form questionnaire from a mandatory census to a voluntary survey.
Stephen Harper knows the collected information will be more than useless. It will be dangerous. Because it will produce misinformation. That’s because it won’t be a full picture. The emphasis on voluntary response means less follow-up, and higher non-response rates from all the people who are too busy or too worn out to reply: immigrants; aboriginals; the ill, the disabled, the aged; low-income households; students and other young people; time-stressed young families; the bankers and the bosses too.
The Prime Minister knows all this, and yet he has stubbornly stuck to his decision despite this knowledge….perhaps because of it.
It’s a decision that can and must be reversed.
The timing is tight but it’s not impossible to get Census 2011 mostly right. Here’s what the next government would have to do:
1) Direct Statistics Canada to undertake the usual number of follow-ups to the 2011 long-form questionnaire, to insure the highest possible response rate in the coming weeks.
2) Conduct a comprehensive evaluation of Census 2011 results within 12 months, to identify where the worst non-response biases are, and attempt to correct them.
3) Provide Statistics Canada with sufficient additional revenues for these undertakings, in order to avoid cutting back on other surveys.
4) Reinstate the mandatory long form census questionnaire
We give governments power so that they serve our interests. On May 2nd, we need to make sure we count, and not just at election time.
Postcript: You can download your very own “I Like It Long!†bumper sticker, courtesy of the Save The Census Coalition.
Given the extra money being spent on the current voluntary long form and the extra money you mentioned that could be used to do some sort of an over sampling ‘ post hoc ‘ in high non-response areas. (I am thinking that violates a few sampling laws, as one needs to draw the sample at the same time, ( I have been down this road!)), so adding these two costs, could we not just delay the long form. I am sure plans are in place, but given the extra sampling costs and non- response damage and then fixing it, I would think a delay is better, say by 6 months. Draw up a new sample using the old 20% method, change the voluntary, and spend extra on an education and awareness campaign to try and eradicate the Harper damage done to the national survey culture. I think it might be cheaper and also be more reliable.
Of Course I do not have all of the info on timing and resources, so it may not be feasible to delay.
When all the history books are finally written, we usually discover why a policy was made. But the decision to end the long form census and the cancellation of PALS may be one set of decisions impervious to historical detective work or archaeology. As Armine points out, Harper benefits from the long form and what they are going to do will be more expensive and far less reliable.
Libertarians were left scratching their heads and my father, an amateur genaeologist stopped donating to the Conservatives for the first time in living memory.
We are all the poorer for the decision and as I tried to point out in an article for the Mark: http://www.themarknews.com/articles/2039-a-country-founded-on-statistics
we will be less knowledgeable about our fellow Canadians than we were on July 1, 1867.
Given seasonal fluctuations in employment and such, wouldn’t it be important to keep the survey month consistent with the old Census data? It comes right after tax season for most, so incomes are likely reported more accurately than otherwise. Of course, this doesn’t apply to the self-employed, who don’t have to file a return until later in the summer but still.
It’s an explicitly negative-sum game. Losing the long form hurts Harper, but he’s betting it will hurt his enemies even more. As to the collateral damage to the country in general, he doesn’t care about that. His politics is, in the end, strictly a struggle for dominance. If he can impair the political forces tending to oppose him more than he impairs his own, that’s a net win, and externalities be damned.
It’s my impression that the right tends to find eliminating information, knowledge and education a net win in that sense. This is just a particularly egregious example.
I believe respondents are asked for their income data and is interpreted as tax data. The item non-response is imputed using Tax data. Mass imputation is also using tax data.
BUt changing the date will have an impact for sure, but I do think it would be less than the massive non-response from the voluntary. Maybe a year delay is more feasible. I do think some kind of plan B is in the works and timing rather than a follow up survey is the way to go on this. The survey Frame from the short form does suffer some decay the more one delays, but it still does not effect it as much as mass non-response.
A study could be conducted to examine these aspects, and I hope somewhere at Statcan they are planning such, in the event of a Tory fail on MAy 2.