Beyond The Commons

 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

When ministers of the crown tweet

by Aaron Wherry on Wednesday, July 7, 2010 9:05am - 0 Comments

Attempting to defend his stance on the census, Industry Minister Tony Clement takes to the Twitter, proceeds to run into economist Stephen Gordon.

TonyClement_MP @wicary Actually the Long Form is still around & will be going to more Cdns–minus the state coercion.

Harbles @TonyClement_MP However the statistical randomness is gone making the data skewed.

TonyClement_MP @Harbles Wrong. Statisticians can ensure validity w larger sample size

stephenfgordon @TonyClement_MP Wrong. Large samples can’t fix sample selection biases.

TonyClement_MP @stephenfgordon Which is why proper weighting will be used, as always the case

stephenfgordon @TonyClement_MP Where will the weights come from? Other voluntary surveys get their weights from the census.

c_9 @TonyClement_MP But weighting is done based on CENSUS DATA. Can’t weight the original data. Answers to these concerns somewhere?

TonyClement_MP @c_9 Folks! There is STILL a mandatory Census!

stephenfgordon @TonyClement_MP How can we reweight for education and income using short form data?

Tags: , ,
Bookmark and Share
  • Andrew (not P or C)

    Well deserved.

  • http://intensedebate.com/people/Harbles Harbles

    Wow I've been published! lol

    That Culture war thing? I think it's on.

    • http://intensedebate.com/people/c_9 c_9

      Me too! Isn't it exciting to be famous? :-)

      This could have been handled much better, but someone appears to have decided that turning StatsCan into some fascist government coercing honest citizens into something they shouldn't have to do made more sense vote-wise. Dreadfully unfortunately, but typical anti-science behaviour common in populist government today (in many countries).

      I disagree with Clement's politics, but do like him as a person and think he's smarter than this. Or thought it anyway – we'll see how the next week plays out.

      • http://intensedebate.com/people/Stewart_Smith Stewart_Smith

        Clement is plenty smart. This is a very smart move for the Conservatives… in terms of votes it means nothing. (In spite of the considerable public clout of u2+gordon.) It will make money for the Conservatives. In the longer term, for a certain subset of their core it will be an important rallying cry once it is fixed by the Liberals. So Clement is smart; ethical not so clear.

        • http://intensedebate.com/people/c_9 c_9

          Good point. I think you've hit the nail on the head.

      • Lord Kitchener's Own

        "someone appears to have decided that turning StatsCan into some fascist government coercing honest citizens into something they shouldn't have to do made more sense vote-wise".

        It could be worse. Down in the U.S., the right-wingers have people so riled up about the evil Census that Census workers have actually been shot at!!!

        • McC

          Wasn't a U.S. census worker actually lynched a couple of years back?

          • http://intensedebate.com/people/charlesh Charles H.

            There was one about a year (or a bit less) back who committed suicide and attempted to make it look like murder (in that the insurance paid extra if he was killed as part of his job), but no lynchings that I know of.

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

          I heard that some wacko killed a census worker and ate his liver with fava beans.

          • RayK

            And a nice chianti.

          • Lord Kitchener's Own

            True, but I think that particular case was more "pro liver" than "anti census".

        • http://intensedebate.com/people/s_c_f s_c_f

          Shot at? Not true.

      • http://intensedebate.com/people/hardmouth hardmouth

        You guys rock! You really handed his crap right back to him! Getting rid of the long form is totally messed up. I think it might actually be the worst thing the Harper government has done so far (which is saying a lot). I always try and refrain from calling this government 'fascists'… but like they're fascists. Canceling all the social/economic data of the census is just appalling. (not to mention burying it the Canada Gazette during the G20, right before July.. hoping nobody would notice. what turkeys!)

        I'm also sort of shocked that the Industry Minister doesn't understand how statistics work. I feel like he must know, or must have been told and is playing dumb, because they've been discussing and fighting with the public service over this decision since Christmas.

        • Lord Kitchener's Own

          I feel like he must know, or must have been told and is playing dumb…

          I wouldn't be so confident that it's an act if I were you. I think he may actually be an idiot.

    • http://intensedebate.com/people/s_c_f s_c_f

      You call that a culture war?

      It's just a dumb decision, nothing more.

      • http://intensedebate.com/people/Harbles Harbles

        But the decision results in the weak and under privileged in Canadian Society being under represented in the Census Data. Who's political agenda does that promote?

  • Matlock

    Is it too much to ask that the minister in charge of StatsCan have a rudimentary understanding of statistics?

    I'll say this much, if he actually took a statistics class when he was at the University of Toronto, they had better fire that instructor.

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

      Totally agree with the first statement.

      I'm not sure about historical U of T requirements, and perhaps someone else knows better than I do, but for awhile, only specialists were required to take a "Statistics for Social Sciences" course, those with a major or minor in Political Science, Sociology, Psychology, (et cetera) were not.

      Also, most of the time, quantitative methods, including sampling methodology, weren't covered in any depth in that course; they would have been, however, in the more advanced Quantitative Methods course.

    • Lord Kitchener's Own

      Is it too much to ask that the minister in charge of StatsCan have a rudimentary understanding of statistics?

      Next you'll be asking that the head of our intelligence service display intelligence.

      You're definitley asking too much.

      Keep in mind that Clement comes from a proud Ontario tradition. At one point while Clement's party was in power here, I kid you not, we had a golf pro for a Premier, a used car salesman as our Minister of Transportation, and a high school dropout as our Minister of Education, all at the same time!

      • Wascally Wabbit

        And don't forget – two doofuses (one being Clement – and one woman who wanted to link Health Privacy legislation to Anti-gang legislation) running Health——-> Downwards!!

      • http://intensedebate.com/people/Stewart_Smith Stewart_Smith

        I had forgotten about the late, great "Any Palladini is a pal of mine". Those really were interesting times.

  • http://intensedebate.com/people/Ottawa_Centrist Ottawa_Centrist

    Mandatory long-form census sheet completion NOW!

    Such a hard topic to rally around, yet a very important one…

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

    This is turning into quite an interesting little issue. It's one of those unique shifts where a party plays to some part of its base on an issue that is one of those longstanding grievances for the base on something that seems pretty small and innocuous. But then is surprised by how many people from the rest of its base reacts because of how they are actually directly impacted by the decision.

    Obviously not enough to turn a Harper supporter into an Ignatieff supporter, but there have been several of these kinds of issues lately. I've never seen so many hard core fiscal conservatives so angry with Harper as I have over the G8/G20 billion dollar boondoggle and porkfest. At the end of the session you had several senior Conservative senators speak out publicly against things Harper or the PMO was doing or saying. Everyone supports the crime initiatives but the size costs Harper is downloading to the provinces for his Criminal Code tinkering that will have questionable/uncertain results is causing a lot of conservatives to choke.

    Not enough yet to go somewhere else, but interesting restiveness brought on by Harper's own decisions.

    • http://intensedebate.com/people/OntarioTown OntarioTown

      It's a Republican thing. The right wing decide in the US to attack the census and the right wing in Canada copy.

      Have to come up with a new divisive one slogan issue now and then. Afterall, how often can you recycle to tough on crime crap.

  • RunningGag

    TonyClement_MP @wicary Actually the Long Form is still around & will be going to more Cdns–minus the state coercion.

    …

    TonyClement_MP @c_9 Folks! There is STILL a mandatory Census!

    I'm confused, when its a long form its state coercion; when its a short form its no big deal? I really like Mister Clement but given how hard he folds when things get tough, I'm starting to question his integrity.

    • John D

      Short coercion is fine, like locking innocent people up for 24-48 hours.

  • John D

    I know a good handful of paranoid libertarians, and I've never even heard them complain about the census. What an odd thing for Clement to hitch his wagon to.

    • Standing By

      I'm sure it was the PM, not Clement, who came up with this. We need to always remember that the PM gets most of his political inspiration from US right-wing blogs, and the census has been an issue for the US radical right for some time. It was therefore just a matter of time until Harper picked up on this issue.

    • Dave

      Fly back to 1996, and listen to the Reformers.

  • John D

    Short-form coercion seems fine. Maybe we should just shorten the gun registration form. Problem solved?

  • Anon 001

    There is something wierd about the Harper government in that it attacks everything that is intelligent or sensible. It seems to resent thought.

    It is hard to believe that Harper would regard his base with so much contempt that he believes that they're incapable of understanding or accepting good policy. His first instinct always seems to be to appeal to the worst in people, not the best; to their fears, not their hopes.

    Maybe that's why, no matter how many photo-ops he stages and how much money he blows on ads, he's still stuck in the low-30s. It's almost as if two-thirds of Canadians have just tuned him out completely.

    Luckily, for Harper, just as many Canadians refuse to tune into Iggy, and a draw is a victory for the incumbent.

    • Mike T.

      that's a bit of an overstatement, but sadly not too too much.

      The man knows his most accessible and likely voting bloc, that's for sure.

    • D-R

      It's like they're proud of being ignorant.

      • Lord Kitchener's Own

        Well, it is blissful.

    • http://intensedebate.com/people/Thwim Thwim

      I think you read Harper wrong. His concern is that his base *is* capable of understanding and accepting good policy when they see hard data for it.

      Hence why he needs to get rid of the data before they start looking at his policies.

  • John

    What little respect the crats at Industry Canada had for Mr Clement is now completely gone.

    How humiliating for them, the government and this country.

  • Guy

    Everyone is making the assumption that the information provided on the long form is accurate. First of all, at least 5% of the forms were never returned. Second, many long forms were filled out with the "assistance" of a canvasser. Psychologists will confirm that people will give a different answer (the answer they think the canvasser will want to hear, rather than the truth) in that situation. Lastly, there was a well organized campaign in 2006 which encouraged Canadians to provide incorrect/incomplete information on the census. As many as 100,000 long forms were impacted. StatsCan has no way of knowing which forms or which questions were impacted. Clearly, the mandatory long form was providing completely unreliable information.

    The data necessary to develop government policy already exists in other places and it is current and more accurate. Federal law requires all departments to provide information request by StatsCan, so the mechanism is in place. Provincial governments track a large amount of information using their Vital Statistics offices (information which is updated daily). There is nothing on the long form that can't be found elsewhere (and it's better information!).

    As an example, use the H1N1 crisis last year. Relying on 3 year old population data would have left a lot of people dead. The provinces used their health systems and vital statistics to get the population information they needed.

    Whenever you wonder why any government makes a head scratching policy decision, you now know why. The government and its bureaucracy are relying on the incomplete, incorrect and dated information from the census. Bad information = bad decisions. Scrap the census all together, and work with the provinces to provide this information on a monthly basis. This may cost a few million to set up, but the governments already exchanges mountains of information, so the start up costs will be low. It will also save the several hundred millions of dollars it costs to run a census. It just makes more sense to do it this way.

    • http://intensedebate.com/people/c_9 c_9

      Where exactly do "vital statistics" come from if not the census? Also, "the mandatory long form was providing completely unreliable information" is not proven by your evidence in the first paragraph. Yes, there are flaws. Yes, it sometimes becomes political. Yes, there are known missing data or gaps. But StatsCan actually measures that stuff and can spot likely flaws and issues – such as suddenly having 1% fewer gay couples, rather than more, willing to report. That was one area that was impacted by political posturing re the census in 2006.

      Also, a common misconception is that the census data is static. StatsCan uses the data to extrapolate and predict any number of things (population, immigration/emigration, income, health) across the 5 year span that each census covers. They do not "rely on 3 year old data", although externally it might feel that way.

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

        Vital Statistics are the registrations of birth, death, and marriage that each province keeps.

        • http://intensedebate.com/people/c_9 c_9

          Ah yes of course, thanks. I was taking that to mean "the various statistics", not those specific provincial ones. Please consider my above comment to exclude that first sentence.

      • Guy

        Why "project" when you have the actual data? The census data is dated and projected and incomplete. StatsCan cannot fill in the gaps without making assumptions. Couple this with the given margin of error inherent in all random samples and you have unreliable information. This is what makes it useless. Do you remember the movie Multiplicity? It's something similar. We have better measures of information available and we should use those instead.

        • http://intensedebate.com/people/c_9 c_9

          Can you provide some examples of these better sources of information? I don't doubt that for some things we do actually have better data somewhere. But so much of our data suffers from selection bias which must be corrected against known proportions, which is where the census data comes in.

          I would love a better system than asking everyone to complete a form by law, but I haven't yet heard of a replacement (even for part of the census) that would provide equally accurate and precise data.

          • Guy

            Vital Statistics tracks all birth, deaths, marriages directly. Health Departments track births, deaths, relocation and new residents. Land titles track home sales. Universities and school boards track enrollments and graduation rates. And so on. Much of this information is already shared. It is simply a matter of an investment into better coordinating this information. I'm sure it would cost significantly less that a census every 5 years. On top of that, it would provide timely and precise data, unlike the census, which I have explained above, which is dated and , because it is based on a sample of incorrect/incomplete data, inaccurate.

            Quite simply, much of the information gathered by the long form is of no business to any government. I know, because I have received the long form in each of the last 4 census. The fact that our governments base policy on this data should scare all of us.

    • ahm

      That's kind of like arguing voting is inaccurate and too expensive, so let's just save the money and get rid of it. (which, to be fair, I actually kind of believe)

      • Guy

        An election is cheaper than a census, although, at this rate, over time, not so much. However, if on election night, the ballot boxes were opened and only 10% of the ballots counted and the results declared on that sample, you might feel differently.

        • a-non

          Well I hope this settles once and for all whether his Twitter is being ghostwritten.

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

    Second, many long forms were filled out with the "assistance" of a canvasser. Psychologists will confirm that people will give a different answer (the answer they think the canvasser will want to hear, rather than the truth) in that situation.

    I can vouch for that, I worked the census of 1986 and I spent weeks following up on incomplete and missing long forms. Without disclosing anything confidential, I can tell you that the value of the data I collected was pretty limited because much of it was inaccurate guesswork or pure fiction. Some people don't trust the government and will not co-operate fully no matter what.

    Frankly, I think Harper's decision here is defensible as a political decision. Clement's decision to start bantering about it on Twitter, on the other hand, is pretty poor thinking all the way around.

  • http://intensedebate.com/people/madeyoulook madeyoulook

    I just wish they would say "the data collected is not worth the coercion because we are a free country" and kill the long form completely. For the quality of the data they're going to generate from the existing long form plan, they may as well.

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

      I'm curious. If statistical data drawn from a voluntary pool is worthless – as many have suggested – then what value is there in commercial polling which is 100% voluntary?

      • http://intensedebate.com/people/c_9 c_9

        I would suggest it's not worthless, but it's not worthwhile at the cost. So much work in public and private sectors is based off trusting census data, so a voluntary pool would be willfully introducing more flaws and bad data which would require more work to work around, and often could not be worked around.

        For purely voluntary polling, it becomes useful by being weighted based on the census data. If you ask 1000 people to reply, and it turns out that 430 of them are NDP voters from large urban centres, you know (hopefully) that this is not a representative sampling. But how do you adjust? The census data (known to be near-factual) helps. If we can't trust the census data – that is, if it goes from 99% accurate to 91% accurate – that becomes a bigger margin of error in everything we do.

        • Guy

          Once again, provinces have population data available. They often release the general information to the public on an annual basis. So that information is current, the census information will be at least 2 years old (and could be up to 7 years old!), since it takes that amount of time for StatsCan to process the information.

          No one should be putting any trust in the census data.

          • http://intensedebate.com/people/ZestyMordant ZestyMordant

            You're right, provinces have population data available. Know where they get it? Statistics Canada. Like these quarterly population reports put out by the Alberta government. Note the citation at the bottom of each table.
            http://www.finance.alberta.ca/aboutalberta/popula…

          • http://intensedebate.com/people/ZestyMordant ZestyMordant

            There are also population totals released every year from Alberta Municipal Affairs that total the populations in each community and municipal district. The population for each community and municipal district are used to distribute provincial and federal per capita grants.

            In Alberta, municipalities have the option to conduct a municipal census and if the minister grants permission they may update their official population. If they don't conduct their own census, population numbers from the last federal census are used. Saskatchewan, on the other hand, does not allow municipalities to update their own population and all municipalities use federal census info.

          • riley

            Municipalities get most of their planning data from the census, as do many provincial governments. I used to work in marketing for a farm publisher, we used provincial ag data that was based on data from the census. Also, the census is almost the only source of information about aboriginals. If Harper's plan is to "disappear, the poor, Aboriginals and the super rich and turn everyone into a middle class Calgary suburbanite, his plan to chop the mandatory long form is a good one.

      • http://intensedebate.com/people/madeyoulook madeyoulook

        "Voluntary polling" of a couple thousand (or so) Canadians over three consecutive evenings is cheap. But the quality is iffy enough over the simple question of "I am a likely voter and if an election were held today I would vote X." It will fall apart completely when you start asking dozens of questions of a personal nature, and just adding more opportunities for people to volunteer does not rescue the data from that flaw.

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

        There's a difference between opt-out polling and opt-in polling, methodologically speaking. Telephone polls, randomly sampled, are opt-out. You hang up, swear at the interviewer, or just don't pick up the phone (try to avoid the second, these folks don't make much money and aren't really the ones with whom you have a beef.)

        Online polls and mail-back surveys are opt-in. You are sent the invitation or survey, and asked to fill it out and submit it.

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

        (continued)

        The principle is that, all other things being equal, you get a more representative sample from opt-out polling than you would from opt-in polling (this bears out in just about every single opinion poll conducted). However, it's not perfect (samples rarely are) so you have to weight it, and the best way to do that is using Census, or 100% sample, data. With the advent of large, highly profiled online panels and methods of random selection within a representative group, the opt-in method of online surveying becomes more sound, but there is some evidence to indicate that not all panels are created equal…at least not yet.

        • http://intensedebate.com/people/novagardener novagardener

          A couple of times i responsed to telephone polls (non-political) but after 7-10 mins I found the questions biased in favour of what they wanted the results to be. Said sorry, will not answer any more questions Neither hubby nor I will respond to any telephone polls.

  • Lord Kitchener's Own

    I only just got the double entendre of the headline.

    Reading Clement's tweets does sometimes feel like being attacked by an animal.

  • hosertohoosier

    Clement is dead wrong on the census, but it is really interesting to see high-ranking officials engage in debate like this (and I'm generally anti-Twitter).

  • http://intensedebate.com/people/psiclone psiclone

    Sorry folks Clement is right!

    • http://intensedebate.com/people/Stewart_Smith Stewart_Smith

      I am sure you mean "right" as in the not a raving socialist sense, (Anyone know why socialists always rave?)

      not in the he is "correct" in the mangling of elementary applied mathermatics sense. (Perhaps he needs a trip to Africa? It would provide an immediate (well 10 month) domestic benefit for our $20 mil, it addition to all the other good stuff that that money will do.)

      • http://intensedebate.com/people/madeyoulook madeyoulook

        I figure he would be in over his head at these new African institutes of learning. At least, if he can't handle elementary statistics, I HOPE he would be in over his head…

  • Mike T.

    Myself and the rest of the Jedi Order of Canada have complete faith in the long form census.

  • Mike T.

    The twitter bit kinda reminds me of when a MacLeans professional blogger got a little hot under the collar at a commenter to the point he kept worrying at it in a later post.

    Is it possible to be "beneath" the fray?

  • Lord Kitchener's Own

    Now that the census paranoia has drifted North, it's only a matter of time before the Tories start telling everyone that Ignatieff is secretly a Muslim who was born in Kenya…

    • http://intensedebate.com/people/sea_n_mountains sea_n_mountains

      i was just thinking, how can things possibly get dumber. i think i have my answer. start your timing devices. it can only be a matter of time….

  • http://intensedebate.com/people/PeteTong PeteTong

    Pride Parade Funding. Fail. G20. Fail. Census. Fail.

  • JamesP

    So how long before we are grossly overpaying a private company to collect data that we would otherwise get from the long form census?

  • madeyoulook

    But I must say I am partial to the idea of elected officials engaging in debate with us mere citizens 140 characters or less at a time. Maybe a leaders' debate can be done this way during the next campaign.

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

    It certainly provides entertainment but I'm not sure there's much upside for the government. Therefore, I would expect to see the practice curtailed in short order.

  • Dee

    This is what you get when politicians, like Tony Clement, don't listen to policy experts: bad policy. Clement's grasp of basic statistics is clearly weak, at best. I guess they didn't teach stats in law school…

  • http://intensedebate.com/people/TwoYen TwoYen

    This entire argument is based on an assumption that people told the truth on previous census forms. How do we know this to be so?

  • Lord Kitchener's Own

    Even if plenty of people lied, where's the evidence that there's any other source of information out there where people are any less likely to lie?

    I don't think anyone's arguing that the census is perfect and infallible, but is there some other vastly superior source of information out there that we could use, or is the argument simply that we should stop even trying to base our policies on reality, because no one can ever get a perfect picture of reality?

  • http://intensedebate.com/people/Stewart_Smith Stewart_Smith

    They actually have to lie in a systematic fashion to screw up the statistics. Conclusions based on census statistics are robust against a limited amount of inaccurate information, but that is only true if there is no selection bias.

Previous

Next

From Macleans