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

Statistical probabilities

by Aaron Wherry on Thursday, March 3, 2011 9:18am - 51 Comments

Pollsters continue to debate the meaning and prominence of their work.

Gregg said the proliferation of sometimes conflicting polls and the hypeventilating analysis that frequently accompanies them does not strengthen democracy. On the contrary, he said: “Rather than have a public that’s informed, you have a public that’s misinformed.” He said he’s not arguing that polls should be ignored; only that their import needs to be interpreted much more cautiously. Rather than pontificate on weekly fluctuations in individual polls, he said it makes more sense to average the results of various surveys and look at the trends over longer periods of time.

It is probably important to consider, as Eric Grenier did this week, how much and how often polling responses change when an election campaign is conducted. Consider, for instance, that the last three changes in government were not obviously foretold by publicly available polling data released immediately before the election was called.

In November 2005, with a few exceptions, the lead for the governing Liberals was variously reported to be between five and nine points. An election was called on November 28 and two months later the Conservatives won the popular vote by six points.

A Gallup poll in August 1993, gave the Liberals a one-point lead over the governing Progressive Conservatives among decided voters, 34% to 33%. As to who would make the best prime minister, Kim Campbell led Jean Chretien by a wide margin, 47% to 24%. An election was called on September 8 and six weeks later Chretien’s Liberals won 41% of the popular vote and 177 seats in the House. Campbell’s Tories, with 16% of the vote, were reduced to two seats.

In June 1984, Gallup gave the governing Liberals a 12-point lead over the Progressive Conservatives, 40% to 28%. The new leader of the Liberal party, John Turner, enjoyed a 20-point lead over Brian Mulroney, 46% to 26%. An election was called on July 4 and two months later Mulroney’s Conservatives won 211 seats with 50% of the popular vote. Turner’s Liberals were reduced to 40 seats with 28% of the popular vote.

(All Gallup data via Carleton’s online library.)

Tags: , , , , , , , , , , , , ,
Bookmark and Share
  • Anon 001

    I think it might be cheaper for all the media organizations and maybe even the political parties to pool their resources together and conduct polls the old fashioned way — door to door sampling. Do it once every month, share the results and everybody will have access to the same data.

    The Canadian public is not that volatile — they don't support Harper 43% one day and 35% the next. It is absurd.

    • McC_

      pooling and sharing? SOCIALISM!

    • ZestyMordant

      Good thought, but there is no way it would be cheaper. In fact, it would be prohibitively expensive. You can't just get a surveyor to go door to door until they get enough responses, because your sample isn't random.

      There are a few ways to overcome this, but they are all very expensive. For example, surveyors could attempt to get a response at every 10th house on each block for every block in the polling area (similar to how the long form census was done until this year). If you're looking at a huge polling area (like Canada), then you either have a huge number of houses to contact or tremendous distances for your surveyors to travel between attempts.

    • Mark

      Pollsters would have trouble getting into apartment buildings, which would pretty much kill the accuracy of polling in urban ridings.

      Plus polling in really rural ridings would take forever seen as how houses aren't clustered in nice little subdivisions.

  • Dot

    I can't help but think that Potter's latest blog is in part a result of the recent poll(s). Had the poll moved in the opposite direction, I doubt he would have been as critical of the Lib's QP strategy (btw I think they should drop it until the Speaker rules on it).

    This was another of Gregg's points – excessive polling drives the media discussion.

    • Wherriggy

      Actually, if you understood the argument Potter was making you would understand that standing in the polls has precisely no bearing on the argument whatsoever.

      • Dot

        If you understood what I was writing, you would know that I was suggesting why HE was writing it at this time was partly due to the polls.

        • Wherriggy

          If you understand the inherently spuriousness nature of non falsifiable claims, you would never have bother to make such a foolish suggestion in the first place, particularly as it has no bearing on the accuracy or inaccuracy of the argument.

          • Wherriggy

            Incidentally, the claim which you claim to be making is also stupid. For instance, Wherry's coverage is typically focused around alleged scandals or controversies (e.g., long form census) which Canadians consistently fail to give a rat's a** about, no matter how many hundreds of posts he racks up about them.

          • Dot

            "^ NOT"

        • Jenn_

          Just In case a nagging doubt creeps into your mind; no, it isn't you. It is Wherriggy

  • Wherriggy

    Keep spinning for your boys, Aaron, I know you're doing the best you can, but you need to try harder: e.g., "It is probably important to consider, as Eric Grenier did this week, how much and how often polling responses change when an election campaign is conducted". Indeed it often does. But it also usually changes in predictable ways – ways in which competent pollsters (as opposed to professional bloviators like Allan Gregg – if there was ever a case of the pot calling the kettle black, it would be Gregg tut-tutting about "pontification") are typically able to predict. Anyway, why don't you do just what your boy Gregg suggests and "average the results of various surveys and look at the trends over longer periods of time"? I think we all know why you don't and won't do that…. because "trends" and "patterns" offer your Liberals even colder comfort than this latest batch of polls.

    • ZestyMordant

      "it also usually changes in predictable ways – ways in which competent pollsters are typically able to predict."

      I'd be very interested to see some examples of this.

    • TimesArrow

      Allen Gregg is a conservative, clown. I believe he's been a pro pollster for more then 20 years now.

      • Wherriggy

        No, he ain't no conservative. You're just such a radical lefty you can't recognize a librull when one is pissing down your leg. Gregg may have worked for a Conservative at one point in the distant past, but sure doesn't make him conservative in any way whatsoever; it was just an expression of his more primordial identity, which is a hack-for-hire. He's published a bunch of b.s. on this that and the other over the years, and you can't find a single conservative utterance in any of it. Lots of pro-Trudeau and various other librull pablum though. (And remember when he was defending the sponsorship program?)

        • TimesArrow

          Wow! A PCer with an open mind. I can see why you despise him. Verboten these days in CPC circles wouldn't you say? Can't blame him at all for not wanting to be associated with the new breed of cons like you. People who can't even write liberal [ librull] without drooling, grimacing and reaching for the crayons to draw horns on any depiction of Trudeau they can stand to lay their hands on. You're a pitiful freak – crawl back under the rock you emerged from.

          • Wherriggy

            Actually, I voted for Trudeau myself, dumbo. Also for Chretien back when Gregg was running Kimmy's campaign. So you might say that Gregg and I have always been on opposite sides, the difference is that I was always right and he was always wrong, and I was acting out principle, whereas Gregg was acting out of his hackish instincts then just as he is now. Regardless, try to find a single conservative (as opposed to Conservative) utterance from Gregg. He's a pure liberal and incompetent to boot.

          • TimesArrow

            "So you might say that Gregg and I have always been on opposite sides, the difference is that I was always right and he was always wrong, and I was acting out principle, whereas Gregg was acting out of his hackish instincts then just as he is now"

            Nice to know you're omniscient. I'll just assume you're right without question now shall i? Consistent though, among the new breed of cons the right to be wrong is a purely liberal characteristic.

            "Actually, I voted for Trudeau myself, dumbo."

            Doubtless you've been beating yourself up for it ever since.

          • Wherriggy

            Now you're just making up random b.s. based on nothing at all besides your own desperate imagination. Why don't you go crawl back into he ***hole you emerged from.

          • TimesArrow

            Fine. Now we've called each other a couple of names we both feel much better.

            Unless you'd like to continue this conversation over at SDA, where i'm sure you'd feel a little more comfortable. Although i wouldn't.

          • Wherriggy

            I have no idea what "SDA" is so evidently you are be comfortable there, wherever that may be, but perhaps you can discuss "it" and other figments of your imagination with your analyst.

          • TimesArrow

            Whatever.

          • Wherriggy

            Burn!

          • TimesArrow

            Anyone who willingly calls himself "Wherriggy" shouldn't go around offering mental health advise.

          • Wherryiggy

            Maybe you're right, but you not what, I didn't offer any "mental health advise" – or even "advice", of the mental health or any other kind. So, in fact, I would say that I am entitled to offer, if not mental health advise/advice, then at least a mental health diagnosis: you are seem to be delusional. Maybe you should look for someone who can advise you about that?

          • TimesArrow

            Have a good life. I mean that sincerely. Life's too short for this bs. I'm sure we'll cross swords again some time.

          • Halo_Override

            Like school on Sunday.

            /Fat Albert

        • Brian

          Allen Gregg is exactly as conservative as Stephen Harper and his deficit-spending, institution-bashing, centralist big-government cabinet is.

  • Wherriggy

    Incidentally, can anyone come up with a series of laments about the allegedly deleterious effects of polling on our democracy that was published back when…. the Liberals were winning? My guess is that, in the minds of these whiners, a simple chage in party status would fix the whole "problem" right there – which means, sadly, that long-term trends indicate that the "problem" will never be "solved".

    • gottabesaid

      Funny, looking back at polls where it showed that the Conservatives were seeing dips in their support, lo and behold, Conservative partisans were discussing how inaccurate polling is, how so-and-so poller is a closet Liberal… now the shoe is on the other foot, and Liberals partisans are discussing how inaccurate polling is. That said, nothing in what I read from this posting or the linked article suggests anything other than the fact that polls are informative, just not 'be all and end all', which is generally how reasonable people with no particular ties to a political party view them.

  • sourstud

    All of this talk about the inaccuracy of polling is ridiculous. The same pundits that criticize polling are the same ones who continue to report on them in the exact same manner as they always have. Rather than suggesting we might be better informed by looking at moving averages over time, why not you know, report the moving averages over time? It doesn't take a statistical genius to calculate these numbers. 5 minutes in Excel would produce the data needed to make a fancy little graph showing trends over time.

    But I suspect Liberal partisans would rather just discredit any poll that shows the CPC gaining traction. It's the same thing from the left every time they see themselves losing support. Rather than up their game, they'd rather change the rules of the game in their favour. The CPC wins government? All of a sudden we need proportional representation. Polling shows the CPC picking up seats? We must change the way polling is done. A media organization wants to start a right-leaning television news channel to counter the CBC? We must pass legislation to block that from happening. The CPC out fund-raises their opponents in television advertising? Find legal loopholes to try to characterize it as illegal.

    It's a tired and pathetic pattern from the Liberals and the rest of the left in this country.

    • gottabesaid

      I'm not going to disagree with your beefs — I do believe the left needs to up their game significantly — but I am going to quibble.

      "The CPC wins government? All of a sudden we need proportional representation." Has this been a Liberal hobbyhorse? Not that I'm aware of. The other left-of-centre parties have been advocating PR for a long time, and whether you disagree with them or not, they make a pretty robust argument .

      "The CPC out fund-raises their opponents in television advertising? Find legal loopholes to try to characterize it as illegal." This is a simplification of this situation. If they're absolved of breaking the regulations, fine. But if they did break the rules, they deserve to be punished.

    • TimesArrow

      . "A media organization wants to start a right-leaning television news channel to counter the CBC? We must pass legislation to block that from happening."

      And the govt tries to remove the rule that the onus is on you to tell the truth at all times.

      You have a point. Too many libs and lefties like to pretend that defending their turf is somehow holier than a conservative defending his. However, let's not pretend the cons don't play for advantage over principle much of the time too.

  • gottabesaid

    "…what we should really be looking at is trends and patterns over time."

    Is there a compilation of this data somewhere? Have you seen this data? I assume that you have since you suggest it "would make the situation look even worse for the Liberals." I don't doubt that you're right, but I'd like to see for myself, if you've got a link or can point me in the right direction. I don't want to be kept in the dark by Wherry and Co.

    • Wherriggy

      It's done and discussed all over the place. There are numerous Wells posts on this subject (specifically, relating to how, for decades, conservatives usually do about five points better at the polls than they do in polls). The most accessible source I know of online which sometimes looks at yearly trends and such like is at places like this:
      http://threehundredeight.blogspot.com

      A lot more information can be found by looking around here:
      http://www.wlu.ca/lispop/

  • notcheapjustfrugal

    Too bad the whole world couldn't be quiet for just one day. Instead of having your head filled with all kinds of nonsense.

  • Andre1958

    Not sure that I agree with everything in the article but I have always said that once the election has been called, all bets are off and everyone in the race starts with 308 seats (except the bloc, of course). Every time the party do/say something stupid, they lose one.

    Due to the large amount of people involved, bias of editors, the proverbial 10 seconds sound byte during the newscast at dinner time and the inexperience of younger/newer candidates, no one is in control and all just hope for the best.

    Of course, it is better to start with high polls but that alone does not guarantee the prize.

    • Thwim

      Hmmm.. interesting theory. Although I think if it were true, we'd wind up with an empty HoC in 3 out of 4 elections.. (19 times in 20)

  • Keith in Brampton

    Once the writ is dropped, no polling should be alowed by anyone other than the political parties, and those polls should be for party use only – no public disclosure.

    We might then have a better chance of having the citizenry pay attention to the actual platforms & issues (and maybe even their local candidates), rather than treating the whole thing as a horse race.

    A side effect would be that reporters would also actually have to focus on the campaigns, rather than filling their pages / airways / web space with vapid analyses of poll after poll.

    • alfanerd

      This is the progressive mind at its best – perceive a slight problem, call for regressive measures in order to fix the slight problem, without considering the consequences of these measures.

      Its not that I particularly care for polls either, but really, if you want to "have the citizenry pay attention to the actual platforms & issues", you're going to have to do a whole lot better then preventing polling during election periods. Besides, if people want polls, let them have it for god's sake… what else are you going to ban in order to get the citizenry to have a debate which is up to your standards?

      It's ridiculous.

  • Olivier

    This polling business is starting to get ridiculous.

  • Mr Irrelevant

    OK so what lessons to draw from what Wherry brings up here? Look at the long-term polling trends as Gregg suggests and you can see the Turner/Campbell pre-election leads were built on nothing at all, and aside from a change in leadership blip, their collapse was, if not predictable, at least unsurprising. As for the third example, the Tories and Liberals had traded small leads over the past year, with the Liberals generally holding the advantage, but a genuine toss-up situation, and then they ran one of the worst campaigns in Canadian history. Long-term trends before elections that were won by the incumbent government were stable leads for the incumbent government, with the possible exception of 1988.

    The long-term trend here is a very firm Conservative lead, ranging from small to large. If an election happens in May, it's hard to see the Conservatives losing without a major event changing the landscape. I don't think I'm saying anything we don't already know.

  • Placentia Bay Ex-Pat

    Election day is all that matters and i can only pray that the day after iggy's on the grey hound back to harvard,then comes boob rae and the liberals slide further down the hole to no where.Well i guess once boob rae takes over he will make quick work of the liberals anyway and he will then drag the dippers down with him and that will ensure a Conservative goverment for many years to come.Lets have that election asap so we can see this happen sooner rather than later.

  • LoyalSubject

    In the context of current polling trends it would appear that Aaron Wherry is dreaming of a miracle that will somehow put his beloved Iggy in the seat of power. Dream on . . .

  • hosertohoosier

    If you actually read the Grenier article, it doesn't simply say that the writ changes things, but rather it suggests that historically the Tories have benefited from such bounces. In every single case since at least 1979 it has done so to the benefit of the right wing party (so the PCs, Reform/Alliance 1993-2000 and the CPC thereafter), with support moving away from the Liberals. Moreover, we have seen time and time again, that as election speculation rises, so does Tory support.

    Why is this? It is possible that Liberal support is, by nature, rather soft – a default option. Alternately, it is possible that pollsters systematically under-poll Conservatives.

  • Loraine Lamontagne

    Polling voting intentions is not terribly interesting to me.

    However, polling of opinions on a precise subject, for example, on scrapping public healthcare, is a lot more interesting.

  • Stewart_Smith

    One idea would be the requirement that methodology and detailed results be released for any poll that is going to be released. (I know that previous statement sounds absurd, but absurdity is the norm in poll-land.) This would make the life of those that aggregate polls easier and allow the wild swings to be damped out. (provided you know who to listen to)

  • Wherriggy

    "Rather than pontificate on weekly fluctuations in individual polls"…. instead, just pontificate weekly, right Allan!

  • Brian

    I agree – although there's one more caveat there, too. Issue polls are as loaded as anything else

    I don't think saying so means pollsters are bad people; I just think we need to keep their results in perspective, in much the same way no one would say doctors have a useless profession, yet no one would say they're all-seeing or infallible either.

  • M_A_D_world

    It might also be nice if the refused to respond numbers were added. If 3 out of 10 hung up the phone without answering the poll, the numbers should reflect that in some way.
    ( Call it the the too pissed to answer these polls group.)

  • Stewart_Smith

    I agree that is perhaps the most important part of any poll, in addition the concept of eliminating undecideds by allocating them between options is bogus.

    I suspect the reality is that a "coalition" of too pissed off & not decided could cruise to an easy majority.

Previous

Next

From Macleans