Beyond The Commons

Aaron Wherry covers all the goings-on in and around Parliament Hill. Follow Aaron on Twitter: @aaronwherry

Categories: Beyond The Commons, Capital Read

The least exciting Mr. Men character

by Aaron Wherry on Tuesday, July 13, 2010 11:30am - 0 Comments

Mr. Census is displeased.

Doug Norris was director general of social and demographic statistics until 2005 — earning the nickname “Mr. Census” — and following his 30-year career atStatistics Canada, he became chief demographer and senior vice-president at Environics Analytics…

He believes privacy concerns around the census are overblown and says the minor hassle of asking one in five households to take 45 minutes to fill out the long form is a fair exchange for gleaning information crucial to Canadian society and the huge range of programs, businesses and non-profits that rely on it. “Moving into a knowledge-based economy, it’s hard to imagine why people don’t value that knowledge and that information much more, and doing away with it at a time when we’re going to need it more than ever is really puzzling to me,” he says.

Tags: census, Doug Norris, Mr. Census, Mr. Men, Statistics Canada

More by Aaron Wherry

BKN Canada

"asking one in five households to take 45 minutes to fill out the long form is a fair exchange for gleaning information crucial to Canadian society and the huge range of programs, businesses and non-profits that rely on it."

Yes, yes, yes, yes, yes, omg yes.

Any libertarian whackjob who has a problem with just should head over to Canadian Tire, buy a rubber raft and some camping gear, paddle out to international waters and drop anchor. And best of luck to you.

http://intensedebate.com/people/lgarvin lgarvin

"Yes, yes, yes, yes, yes, omg yes"

I wish I enjoyed that as much as you did.
Bahamaat

But don't try buying it on your CT credit card or get the extended warranty – they may ask you a few questions, and keep the answer in a database that is a lot less secure than any StatsCan server around.

http://intensedebate.com/people/sourstud sourstud

Yes, but it's voluntary at CT, not forced by law.

Holly Stick

So what?

http://intensedebate.com/people/sourstud sourstud

I refuse to feed the trolls.
Holly Stick

So you don't have an actual argument about why "voluntary" is better than "forced by law"?

http://intensedebate.com/people/LynnTO LynnTO

I agree with his sentiment, but shall elaborate on his current perceived bias: in working for Environics, he can count himself among (one of the many) research firms who use Census data – albeit probably mostly short-form – to weight and compare their research results from public opinion surveys.

Livebloggin Junkie

Lynn, as you are much much more knowledgable on all things statistics then I, maybe you could answer my question. The main complaint from the mandatory Long Form fans is that by making it voluntary it increase selection bias, but couldn't StatsCan use the Short Form to establish the weights for the voluntary Long Form?

http://intensedebate.com/people/tedbetts tedbetts

No because the short form is measuring different things. If, say, wealthy educated born-in Canada citizens have a much higher rate of response on the voluntary form than, say, poor uneducated immigrant citizens, then the short form cannot help you re-weight that, for example.

Livebloggin Junkie

But if the mandatory long form went to 20% of Canadians, didn't they need to weight it against something? It was always a sample and therefore needs to be measured for accuracy against the Census (ie the short form that counts everybody).

Lord Kitchener's Own

The point is that the 20% who get it are a random 20%, and StatsCan is very careful to ensure that it is a random sample. As soon as people get to decide for themselves whether or not to complete the form it becomes self-selected, not random at all. You can extrapolate from a large enough random sample to a larger population, but a voluntary survey does not a random sample make.
http://intensedebate.com/people/tedbetts tedbetts

I'm not a statistician but from what I understand, a 20% sampling – if it is random – is sufficient to extrapolate to a greater population. In other words the variations that one might get between a 20% sampling and a 50% or 100% are within margins of error, 19 times out of 20, etc.

Weighting on the other hand is something you do when there is no random sampling. For example, when you take a poll of voter intentions and the results show that you have a 50% response rate from urban residents and 50% from rural residents. This will result in a much higher percentage of apparent vote support for the NDP and a much much higher apparent support for Conservatives that we know would not be accurate. Weighting is the process employed to adjust the value of the 50% urban responses up to 80% to reflect what we know is the real demographic.

The short form and long form are measuring different things so one can't be used for the weighting of the other.

http://www.calgarygrit.ca CalgaryGrit

You could weight it against gender, location, age, and other short form variables, but that wouldn't eliminate the problem.

Because of low income women and less likely to fill out the Census than high income women, you'd still have bias. Ditto for aboriginals, newer immigrants, etc.

Weighting would only solve the problem if it was a case of certain gender/age groups being less likely to fill out the form.
http://intensedebate.com/people/LynnTO LynnTO

Mostly what ted said. You can't "weight" ethnicity if you don't have an accurate ethnic profile, for example, but I'll go further. Let's say that income is a long-form question, and 90% of those with incomes over 80K respond and only 10% of those with incomes under 20K respond.

a) You'd never know your data were skewed in the first place. Whereas the purpose of the 20% Census is to get "true" distributions, you don't want to ever have to weight the numbers, because that means you're trying to project "true" distributions, rather than just counting them. This is overly simplistic – from time to time Census numbers do require adjustment – but the general principle applies.

Livebloggin Junkie

THANK YOU!
ted and LKO tried but it was your explantion of true distribution that helped me see that the don't weight the NHS.

http://intensedebate.com/people/LynnTO LynnTO

b) When you weight up a sample by a significant amount – in our example, it would have to be double or potentially triple its initial count – you skew the data, even by virtue of attempting not to skew the data. You make those individuals you weight up responsible for their entire group…and, if you don't have the numbers to justify doing it (at the Census Tract level, generally, you don't) that means you're risking projecting half-true demographics onto an entire population, because the number of permutations you reasonably get in the responses to a long-form census is greater than the number of individuals who respond. Then, you don't have a Census, you have a guess.

http://intensedebate.com/people/Gaunilon Gaunilon

I'd have to agree. The government has the right to demand taxes (when necessary and when spent for the common good), and the government also has the right to demand information (when necessary and when used for the common good). This should be a non-issue, particularly in a country that demands so much more than census forms in the name of the state.
http://intensedebate.com/people/auntie_em_m auntie_em_m

Hard to imagine is right. It raises all sorts of paranoid imaginings as to why Harperites want to do this. Mind you, so much of Harper's world raises hackles that I understand why so many Canadians choose to ignore news like this. I'm sure he weighs that outcome in his favour, especially when designing budget bills, for example. Ugh!
http://intensedebate.com/people/TJCook TJCook

Props for a good headline, Aaron!
Johathan

Hey if you don't like the long form, then just skew the data with random information. Useless data is worse than no data.

"it’s hard to imagine why people don’t value that knowledge and that information much more" People DO value the information, that is why they don't want to be forced to give it away for free. The government can easily resolve the issue by allowing the few people who don't want to fill in the form to opt out.

http://intensedebate.com/people/Crit_Reasoning Crit_Reasoning

Intentionally filling out the census form with false data is a crime, believe it or not.
http://intensedebate.com/people/LynnTO LynnTO

Useless data is worse than no data.

Which is precisely why making it optional is dumber than not having it at all. Don't falsify information, people, if you don't like it, don't respond.

Standing By

At first I thought this change in the census was just a mistake. I am coming to think it was a thought-out strategy, however, and that Harper really does want to set the stage for future ideological-driven social and public policy that can't be countered by recourse to pesky "facts".

This is actually quite astonishing. This guy is a long-term thinker whose will indeed change this country so you can barely recognize it. If this is okay with most Canadians, then they deserve what they will get.

TonyB

Nothing Harper does these days seems to be thought out at all

Lord Kitchener's Own

Some may consider Mr. Census to be the least exciting Mr. Men character, it's true.

Others, however, will invest in thick steel security doors, guard dogs, and firearms, lest the absolutely terrifying Mr. Census ever approach their home. He's like their Freddy Kruger. Only more evil, and less funny.

http://intensedebate.com/people/lgarvin lgarvin

I've always been afraid of Mr. Rooter.

http://intensedebate.com/people/LynnTO LynnTO

You should hang out with Mr. Transmission, he's a friend of mine.

http://intensedebate.com/people/lgarvin lgarvin

Mr. Muffler told me something pretty disturbing about Mr. Transmission.

But he swore me to secrecy.

http://intensedebate.com/people/LynnTO LynnTO

Mr. Rogers would say that's not very neighbourly…
Dave

Mr. Tambourine Man said Mr. Rogers is a — I don't think I can say it on the internet.

It's really dirty.

John

Facts and data have a Liberal bias.
http://intensedebate.com/people/E_B_ E_B_

I figure that the people in my home town have been doing everything to mislead or disrupt the census for the last 10 years. Our population has tripled in that time, and we still don't have a single 4 lane road in/out of town completed yet. I don't know how the heck we are supposed to commute.

If we expect the government to provide basic services and infrastructure, doesn't it make sense to collect data?

I wonder if that group the Conservatives are pandering to this time only pay cash for all their purchases. Their heads would surely explode if they knew how much information credit card companies and loyalty programs had about them.

http://intensedebate.com/people/LynnTO LynnTO

Maybe. I'm more inclined to believe that fast-growing municipalities just aren't billing enough in development charges to cover the roads (et cetera) needed to connect people to services they require, and are as a result forced to use their infrastructure budget – that would otherwise go to expanding existing thoroughfares – to cover the shortfall.