Wednesday, April 20, 2011

Parties hold firm

With only today's Nanos poll to add to the projection, there hasn't been much change. In fact, there has been virtually no change at the national level. But a few new seats are getting close to the tipping point.
The Conservatives and Liberals are unchanged at 38.7% and 28%, respectively, while they are still projected to win 147 and 80 seats. The New Democrats are up a tiny 0.1 point to 17.9%, and remain at 35 seats. The Bloc Québécois and Greens are unchanged as well, at 8.6% and 5.8% respectively. The Bloc is also projected to win 45 seats.

Regionally there have been a few more changes. The Conservatives were stable in most parts of the country, but also gained 1.1 points in the Prairies. They are now at 51.7% in Saskatchewan and Manitoba, better than their 2008 performance.

The Liberals dropped 0.3 points at both ends of the country, but continued their slow, incremental progression in Ontario. They now stand at 35.5% there.

The New Democrats had more movement, with gains of 0.3 points in British Columbia and Alberta, and gains of 0.2 points and 0.7 points in Ontario and Atlantic Canada, respectively.

None of this has resulted in any seat changes, however. But a few seats that have not changed hands so far in the campaign in the projection are getting close.

The Liberals are within one point of the leader in three ridings: Vaughan (46.1% Conservative to 45.2% Liberal), Brome - Missisquoi (31.1% Bloc to 30.7% Liberal), and Saint John (43% Conservative to 42.3% Liberal).

The New Democrats, meanwhile, are within one point of the leading party in one riding: Vancouver - Kingsway (35.2% Liberal to 34.4% NDP).

But things are still fluid, as the Conservatives are also still very close in some of the ridings that have recently switched over, like Brampton - Springdale, Kitchener - Waterloo, and Sault Ste. Marie. Depending on how the polls move in the next few days, we could see the Tories back over 150 or below 143, the amount of seats they held when the election was called.

On an unrelated topic,'s newest sponsor, 270soft, designs election simulation computer games. They've recently released their version for the 2011 Canadian election. It looks like a lot of fun, and I've been reliably informed that some of their older titles were great. So, I invite you to check them out.


  1. Is there any way to account for intangibles here? As an example, the Conservative numbers for the two Thunder Bay ridings have not changed - one's even gone down. However the fact that the PM held an evening rally and then morning events has to mean that they see the ridings as potentially winnable. Currently you have the Thunder Bay Rainy River CPC number at less than their 3rd place finish in 2008.

  2. Just a general comment on your site - I see you break the poll results down by region, but any chance of a breakdown between Quebec and the Rest of Canada?

    It seems that the real game-changer this election is the NDP surge in Quebec . . . but at the same time, there's no corresponding surge for them nationally. Some quick calculations based on the latest Nanos poll suggests to me that their support outside of Quebec has dropped from 20.1% in the 2008 election to 18.1% today. Any comments?

    Thanks, and thanks for this great resource.

  3. Gitfiddler,

    The model needs to be applied uniformly for it to have any rhyme or reason, so I avoid these kinds of subjective tweaks.

    If a poll is released for those individual ridings, then I would have something to work with. So far, we've only see riding polls for a few ridings in Quebec.

  4. I understand that your projection is designed to avoid big fluctuations caused by individual polls. But seriously, what will have to happen for your projection to show the NDP above 20% in Quebec?

  5. Anonymous 08:47 (#1),

    Yes, it does seem that much of the NDP boost has come in Quebec - if not all of it. It gives a bit of a misleading impression, as the NDP will have a lot more trouble turning that gain into seats than they would if the gain was coming in, say, BC or Ontario.

    Anonymous 08:47 (#2),

    Just a few more days of polls like this. But the Harris-Decima poll putting the NDP at 15% in Quebec put a damper on their progression.

  6. What a complete waste of time and money this election will have been if we come out the other end with the same distribution of seats. Perhaps if voters reflect on THIS possible outcome they might give themselves a good shake. There is still time . . .

  7. @Dio

    As unfortunate as that would be, the $300-350m for the election is a small amount compared to the total cost of ANY party's platform. The waste of time and money for the election should be well down the list for anyone's voting priorities.

  8. An awful lot of polls show the NDP in Quebec at mid twenty levels. I look forward to the next CROP poll which breaks down Quebec support into regions. If the NDP are to translate this unprecedented level of support into seats, their vote will have to also be efficient, meaning they would have to be gaining more traction in specific areas rather than across the board.

  9. The problem is voters don't pick "the government". Voters pick their local representative. I'm sure Liberals AND Conservatives want the seat distribution to be different from last time, to say nothing of the NDP and Bloc. But it's not different, because the country isn't different.

    Conservatives are still terrified that Canada has become a "failed Northern European welfare state".

    60% of the country is still terrified of a Harper majority. But despite how people say "the left" is fluid, it ignores that there are legitimate differences between the parties that voters can't stomach. The Separatists aren't gonna vote in the party that brought them centralized corruption. A unionist isn't gonna vote in the party that brought them NAFTA. *Maybe* you'll see Green voters go Liberal.

    Yeah, the Liberal (and liberal) panic will generate a lot of strategic anti-Harper votes. But the 100+ races where it's Liberal vs. Conservative will stay that way regardless of "strategic voting". The big result is what it will cost the NDP for the 70+ races in the country that are NDP vs Conservative and the 30+ races that are NDP vs Liberal.

    Quebec is the wildcard this election. If there's going to be any significant movement, it will be because of Quebec.

  10. Looking at your charts below, it appears that the Quebec figures are weighted based on pre-election polls (19%), pre-debate polls (45%) and post-debate polls (36%). In response to the NDP surge in Quebec, you may wish to re-evaluate your weightings.

  11. Hey Eric, presuming Anonymous 10:12 is correct, and CROP does release regional Quebec polls, will you be able to incorporate them on a regional basis to your system? It seems like how Quebec turns is turning out to be *the* issue here, and this may be the best opportunity to get a more accurate picture of this, since we're not realistically going to get 75 riding polls...

  12. Anonymous 10:30,

    Patience, patience. If the trend continues it will be reflected in the projection. This is, really, a very new state of affairs. We have to wait and see whether it endures.


    No, the model is not calibrated in that way. We are seeing some riding polls coming out in Quebec, however, so we're not running completely blind. In fact, polls for the two northern Quebec ridings were released today.

  13. Eric

    I saw on CTV that Jean Lapierre indicated that the CPC will hang on to all their Quebec seats.

    He made a passing reference to some Quebec City polls.

    They I came to your site and found that you have the CPC retaining all their Quebec seat despite the overall CPC Quebec numbers dropping like stone to 15%

    Look like your model is robust.

  14. Could you please sort your tables at the bottom of the page by either date or weight, so that they are more comprehensible? Many thanks.

  15. Wouldn't it be reasonable to enter the results of regional polling as weak-weighted "riding polls" or something, so that the information still did get passed on to the model?

  16. Anonymous 10:47,

    I realize that would be easier to read, but I cannot do that.


    It would be, and perhaps in the future I will have some mechanism for it. I don't for this election, however, and an additional problem is that the regions included in the polls are never well-defined.

  17. what % of the votes does Harper need for a majority?

    In 2008 he got 37.63.

    If they would have got 30,000 extra votes (CPC supporters that stayed at home) in 20 close ridings they would have had 163 seats with 37.78% of the vote.

    I know this is just fun with numbers but it shows how hard it is to translate national or even provincial polls to a riding level.

    The Liberals were a tied or ahead in the polls in Ontario when the Vaughn By election had a 17% point shift from Liberal to CPC.

    These numbers also show why the CPC are focused on 30-50 ridings. These all could be comfortable CPC wins and not show up at all in the national/provincial polls.

  18. Eric,

    At what point, percentage wise, do you think the NDP stop becoming a spoiler/splitting vote in Quebec and starts taking ridings away from the other parties, particularly the Bloc, without handing close races to the Liberals or Conservatives?

  19. There's something fishy in today's Nanos. For the Liberals, how does a 8.2% drop in BC lead to a national drop of 1.8%? For the NDP, the regionals should only add up to a 1.5% rise, not 2.5%. Either the national numbers from today or yesterday are wrong.

  20. I think it is unavoidable that both happen. But probably if they actually get 25% or so.

  21. "These numbers also show why the CPC are focused on 30-50 ridings. These all could be comfortable CPC wins and not show up at all in the national/provincial polls. "

    Oh Dear !! Into the BC Gold again I see. SAD !!

  22. @BCVoR

    "These numbers also show why the CPC are focused on 30-50 ridings. These all could be comfortable CPC wins and not show up at all in the national/provincial polls."

    And they could all be routs of the CPC, and that also would not show up in those polls.

    Or, Éric's model could be good and the ridings that he has as close actually ARE close.

    We can guess right now, but it remains a guess (or maybe wishful thinking) with no factual basis to back it up. Only way to know for sure is to wait for May 2. Until then, the reasonable way to deal with it is to assume that they won't swing to one extreme or the other - as that is, historically, the most likely situation.

  23. I can say with a great degree of confidence, the Conservatives are very happy with the numbers they are seeing, given their behaviour.

    The Liberals are not. When the Liberals are campaigning now and plan to campaign next week with Jean Chretien holding Iggy's hand in fortress Toronto, they are in trouble.

    The NDP don't have the ground game in Quebec outside of a few riding to bring in the vote. None of the pundits, including Chantel Hebert believe the vote numbers will be as high as their poll numbers.

    There is also serious concern about the Liberal ground game in most of the country.

  24. I just did the calculations, I was wrong earlier.
    However, I did discover that Nanos hasn't been weighting the provinces by actual population. For example: Ontario 38.7% of Canada (actual) 30% (Nanos survey). And the Atlantic provinces & the West are oversampled in his survey (46% of Canada instead of 37.7%)

  25. Yes, but I'm sure that Nanos weights those regional disproportionalities to get his national %s. That is standard practice.

  26. Brendan .... EKOS always over polls Quebec.

    I think that they normalize the numbers to get the national numbers

  27. @Brendan Kane

    Nanos weights geographically, so that while for example Ontario is only 30% or so of the sample, theye are weighted back up to ab 38.7% in the national toplines...

    Using todays poll for the Tories....

    Nanos% Pop%
    Atlantic 0.382 * 0.07 => 0.02674
    Quebec 0.168 * 0.232 => 0.038976
    Ontario 0.449 * 0.387 => 0.173763
    Praries 0.553 * 0.177 => 0.097881
    BC 0.418 * 0.133 => 0.055594
    National 0.392954

    The 39.29%, give or take a little rounding error, is the 39.1 he reports today

    The Vorlon

  28. Here's what I got for the NDP. Are my calculations correct?
    Atlantic 0.255 x 0.070 = 0.01785
    Quebec 0.254 x 0.232 = 0.058928
    Ontario 0.127 x 0.387 = 0.049149
    Prairies 0.168 x 0.177 = 0.029736
    BC 0.247 x 0.133 = 0.032851
    National 0.188514

  29. @ brendan kane

    yes, your numbers look correct, I think NANOs may have had a mild miscalculation in his poll....

    Maybe it was supposed to be 18.9% and the 8 and the 9 just got transposed?

  30. @ Brendan Kane,

    I have exchanged a couple emails with Nik over at Nanos and we are entirely on the wrong track here.. the calculations we are doing do NOT apply.

    The Nanos poll is stratified by age and NOT geographically, so the fact the Tory numbers happened to work out was, well, just a fluke...

    the Vorlon


COMMENT MODERATION POLICY - Please be respectful when commenting. If choosing to remain anonymous, please sign your comment with some sort of pseudonym to avoid confusion. Please do not use any derogatory terms for fellow commenters, parties, or politicians. Inflammatory and overly partisan comments will not be posted. PLEASE KEEP DISCUSSION ON TOPIC.