Wednesday, March 2, 2011

Liberals projected to lose seats with new projection model

Well, it took some time to complete but it is finally done. The new projection model is up and running and ThreeHundredEight is now able to make riding-level projections for all 308 ridings in the country. When there is less going on, hopefully this week or next, I will post a full methodology with test results for the new projection model. But in the meantime, with polls showing such a huge shift in voting intentions, I couldn't let the projection of February 18 stand.

Note that you will always be able to click on the links to the right to see the riding-level projections. You can also see them at the bottom of this post.

ThreeHundredEight now projects if an election were held today that, with 38.1% of the vote, the Conservatives would win 152 seats, a gain of eight seats over the last projection of February 18, and a gain of nine seats over their current standing in the House of Commons.

With 27.7% of the vote, the Liberals are projected to win 73 seats, a drop of 19 from the last projection and a drop of four from their current crop of MPs in Ottawa.

With 9.9% of the national vote, the Bloc Québécois is projected to win 52 seats in Quebec, unchanged from February 18 but a gain of five compared to their current caucus.

With 15.5% of the vote, the New Democrats are projected to win 31 seats, a gain of 11 over the last projection but a loss of five overall.

With 7.4% of the vote, the Greens are not projected to win any seats, and no independents are projected to win.

It is very important to note that a lot of these gains and losses are due to the new projection model. The old projection model would have projected 148 Conservative, 86 Liberal, 52 Bloc, and 22 New Democratic seats. In other words, four of the Conservative gains are due to the model, as are nine of the NDP gains. Thirteen of the Liberal seat losses are due to the model change. The old model would have projected gains of four Conservative and two NDP seats, and a loss of six seats for the Liberals.

Obviously, the old model was harsh on the NDP to the benefit of the Liberals. It was not able to simulate accurately the regional strength of the NDP within each province. However, that is not to say that the old projection model was wrong. While I am more confident in the new model, in the end we will have to wait and see when the votes are counted whether the old or new model was more accurate.

Nationally, the Conservatives are up 1.1 points and the NDP 0.4 points. The Liberals are down 0.4 points, the Bloc is down 0.1 points, and the Greens are down 1.1 points.

In British Columbia, the Conservatives have gained 1.9 points and lead with 40.3%. They are projected to win 22 seats, unchanged from February 18th. The Liberals are down 2.3 points to 23.7%, and three seats to seven. The NDP is up two points to 23.2% and three seats to seven. The Greens are down 1.7 points to 10.4%.

In Alberta, the Conservatives are up 0.6 points to 61.4% and are now projected to win all 28 seats. The Liberals are up one point to 19.9%, and lose the seat they were projected to win. They are still very close in Edmonton Centre, however. The NDP is down 1.3 points to 9.4%, while the Greens are down 0.3 points to 7.5%.

In the Prairies, the Conservatives have gained 4.9 points and lead with 51.5%. They are projected to win 22 seats, up one. The Liberals are down 2.8 points to 21.2%, followed by the NDP at 21.1% (+0.9). The NDP is projected to win four seats (+2) and the Liberals two (-3). The Greens are down 1.9 points to 5.6%.

In Ontario, the Conservatives are up 0.8 points to 41.2%, and are projected to win 56 seats, a gain of two. The Liberals are down 0.6 points to 34.6% and are projected to win 35 seats, down seven. The NDP is up 1.1 points to 15.3%, and is projected to win 15 seats, a gain of five. The Green Party is down 1.2 points to 7.8%.

The Bloc Québécois is up 0.4 points to 40.1%, and is still projected to win 52 seats. The Liberals are up 0.9 points to 21.3%, and are projected to win 13 seats, down one. The Conservatives are up 0.2 points to 19.2% and are projected to win nine seats, up one. The NDP is down one point to 12.7%, and is still projected to win one seat. The Greens are down 0.6 points to 5.7%.

In Atlantic Canada, the Conservatives are up 2.8 points to 39%, followed by the Liberals at 36.3% (+0.5). The Conservatives are projected to win 13 seats, a gain of two. The Liberals are projected to win 15 seats, a loss of three. The NDP is down 1.5 points to 18%, but is up one seat to four in the region. The Greens are down 0.3 points to 6%.

In the North, the Conservatives are projected to win Nunavut and Western Arctic, while the Liberals are projected to win Yukon.

At the next projection update, I will identify the individual seats that will have shifted from one party to another.

Here are the full riding projections (click to expand):
Quebec, Atlantic, North


  1. Very cool Eric. The results suck but still very cool.

  2. Yes, I suspect there will be a few gasps when people see the new Liberal number.

  3. Very interesting to see how close some ridings are. Especially in Ontario. Depending who does well in the campaign could see a ton of seat change.

  4. I don't understand. What has changed from the previous projections?

  5. The NDP seat count makes a lot more sense in this projection model. I'm not just saying this as a partisan - I honestly think that the old model tended to consistently low ball the NDP seat count.

  6. Thats quite the massive swing in Avalon. How can you go from a Lib win by 10 points, to such a huge gap in favour of the Cons?

  7. Looks like the Libs have a chance in Edmonton Centre if things improve just a bit for them. A non-compete deal between the Libs & NDP in Edmonton Centre and Edmonton Strathcona could benefit both of them, to the expense of the Cons. While I don't advocate a merger or deals across the country, these narrow, seat-for-seat deals in local areas (where volunteers can be moved around easily) might be something for them to consider.

  8. Anonymous,

    For Atlantic Canada, I am distributing the vote to the four provinces according to the proportions of the recent Angus-Reid poll.

    In that poll, the Conservatives are ahead in Newfoundland & Labrador. I currently have them at 42.1% to the Liberals' 32.7% in the province.

    That's a huge change from the 17% or so support the Conservatives had in 2008, almost entirely because of the ABC campaign.

    Now that the Conservatives are back in it in Newfoundland, the results from 2008 really can't tell us too much.

    But with the Tories at over 40% in Newfoundland & Labrador, Avalon becomes a prime seat for them.

  9. Looking at the riding by riding results for BC, for example, I don't disagree with the final conclusions.

    Nevertheless, some numbers within those conclusions leave me somewhat perplexed.

    Such an example is the riding of New Westminster-Coquitlam.

    Your projected federal vote in BC (with change from 2008 in parentheses):

    CPC: 40.3% (-4.2%)
    Lib: 23.7% (+4.3%)
    NDP: 23.2% (-3%)
    G: 10.4% (+1%)

    Your projected vote for New Westminster-Coquitlam (with change from 2008 in parentheses):

    CPC: 33.3% (-5.5%)
    Lib: 11.2% (+-0%)
    NDP: 51.3% (+9.5%)
    G: 4.2% (-3%)

    So you have the CPC down by 4%, the NDP down by 3%, and the Libs up by 4% in BC overall.

    Yet your projection for New Westminster-Coquitlam shows completely contradictory swings.

    I can't figure that one out.

  10. New Westminster - Coquitlam had a by-election. I took that into account.

    And the middle is not a simple swing model, there are individual factors included for each riding.

  11. I was going to question just how accurate the Independent predictions can be without riding-level polling, but then I remembered that Bill Casey stepped down, and that Helena Guergis and André Arthur are the only two sitting Independents.

    I still think this would under-represent support for specific independent candidates, but I don't think it'd change the results at this time.

  12. Middle = model, in the above post.

  13. wow, quite a result. lots of close races that could be tipped either way in a campaign. i love the detailed riding results. Bravo!

  14. Glad to see that you finally finished your model and that the results are much closer to my model. As I was saying complaining for months about your old model, you were too nice with the Liberals and too hard on the NDP.

    Seems our two models will be very close (as well as DS). I think it was more fun when they weren't tough lol

    Good job Eric.

    Bryan Breguet

  15. Yay, your model's projections are now much closer to mine! The low NDP and high Liberal numbers you had before always made me a bit uneasy. I'm glad to see that we now mostly concur :)

  16. I'd wondered about your old projections giving the NDP so few seats, given that they weren't that far from their 2008 vote totals. This now makes much more sense.

    I'd gotten used to seeing higher Liberal numbers, though, so this is a bit of a shock.

  17. I cannot imagine Sionhan Coady holding onto SJSMP or the CPC winning Random Burin-St. Georges(think thats the name) but we'll see.

    Avalon will be the most exciting riding in NL seeing if Senator Manning runs he supposedly cannot win it.

  18. Eric does your new model include any modifiers for incumbency and for star candidates ?

    I ask because in Burnaby-New Westminster a former MP Paul Forseth is running.

    He used to represent New Westminster – Coquitlam – Burnaby and then after re-districting New Westminster – Coquitlam.

    What's going to be a very close race you have projected to be an NDP blow out.

  19. Very impressive!

    I might have a few quibbles with a riding here or there (though, not many), but I can't quibble with the overall numbers.

  20. As I'll explain when I do the methodology write-up, yes, incumbency and star candidates are some of the factors taken into account.

    I'm open to suggestions of who would be a star candidate. I don't have Forseth listed as one. I'm not sure if a former backbench MP fits the bill.

    But you may attempt to persuade me.

  21. Carl,

    Quibbles are welcomed. It's a work in progress, and sharing local knowledge of individuals ridings will only help.

  22. Since all polls are "yesterday's" polls, they only represent the starting point for any next election.

    Holistically, they may only indicate a momentum trend that can only change because of an election campaign.

    Currently, we must assume that the Conservatives will win an election, and the quality of their election campaign will determine if they win a majority or minority.

    Of course, unexpected events can quickly change the picture as we know only too well!

  23. Eric there's lot of former MPs running (from all parties) so I won't ask you to address Mr. Forseth in particular.

    But perhaps you should look at a way to address them all.

    I'd imagine, by virtue of increased name recognition alone, there would be some difference in their impact vs Joe Random who nobody has ever heard of.

    Then again maybe its all a wash, they could have negatives that got them defeated in the first place.

    Just a suggestion.

  24. It is one of the things I intend to look at. I have a list of factors I need to investigate, but there might not be enough time before the election to get through them all.

  25. "I ask because in Burnaby-New Westminster a former MP Paul Forseth is running.He used to represent New Westminster – Coquitlam – Burnaby and then after re-districting New Westminster – Coquitlam."

    For the zillionth time we do not "redistrict" in Canada - that is what they do every ten years to the US House of Representatives. In canada we REDISTRIBUTE.

    Forseth was regarded as a very weak MP and you may recall that he wanted to run in the byelection last year in NWC and the party basically told him to get lost. There virtually 0% overlap between Burnaby-New Westminster and the riding that Forseth used to represent.

  26. My comments on your BC sests

    Esaquimalt Juan de Fuca - this is a pure NDP Conservative battle with odds in the Conservatives. The Liberals are not issue with the retirement of their MP Keith Martin who won it originally as Reformer.

    BC Southern Interior will be closer than you project, the past numbers are skewed because of the 2006 election.

    Burnaby Douglas will be a lot closer though with the NDP still the most likely winner but the Conservatives having goods because there is no incumbent. Projecting a Conservative win here is not one I would bet on, this riding has its own mind.

    Newton North Delta will be closer and the NDP vote will be lower than you project

    Projecting that the Conservatives will lose North Vancouver seems odd to me.

    Saanich Gulf Islands I agree Conservative win, but higher Liberal vote is realistic. 25-30% is more realistic. I see the NDP doing worse and the Greens coming about tied with the Liberals.

    Surrey North will much closer between NDP and Conservatives.

    Vancouver Kingsway is not likely to be lost by the NDP.

    Vancouver South will be two person race between Libs and Cons with both breaking 40%. It is a re-run for the two parties of 2008 and the Conservatives have more money and people on the ground.

  27. Eric,

    I was wondering if you could comment on the Edmonton Sherwood Park Results that have such a High Other number. I was unaware of anyone running in Edmonton of that prominence as an Independent. Also, I was curious about Simcoe Grey, other showed around 23% if I recall correctly.



  28. DL your facts are wrong.

    "In canada we REDISTRIBUTE."

    Actually according to EC its known as "readjustment of electoral district boundaries".

    However, "redistribution", "redistricting", and "readjustment" can all be be used.

    "There virtually 0% overlap between Burnaby-New Westminster and the riding that Forseth used to represent."

    A quick look at the google maps function over at pundits guide shows that there is about 33% overlap.

  29. Bernard von Schulman here's my take on some of your comments:

    Esaquimalt Juan de Fuca - Agree 100%. CPC-NDP race, CPC edge, LPC in third.

    BC Southern Interior - It'll be closer but edge on NDP. Gun registry issue will hurt the NDP but this is hippy country, lots of bc bud, the shambalah music festival takes place in this riding...

    Burnaby Douglas - Agree NDP will be stronger but I wouldn't give them the edge. 100% toss up.

    Newton North Delta - Disagree. Jinny Sims is well known, will be strong. NDP vote will be higher, LPC vote lower, 3 way toss up.

    North Vancouver - Agree, Andrew Saxton will hold. He's already a parliamentary secretary, seen as a rising star.

    Vancouver Kingsway - Disagree, LPC candidate did well in '08 which was a horrible cycle, is running again.

    Vancouver South - Agree.

  30. Rocky,

    Jimmy Ford is running in Edmonton - Sherwood Park again. He had about 30% of the vote in 2008.

    As for Simcoe - Grey, that is Helena Guergis's riding. She will be running again as an independent.

  31. Congratulations on your new model.

    what for the "Flames" when it hits the G&M

    Liberals Holland to lose by a hair to Alexander in Ajax-pickering and Siobhan Coady losing in St. John South. They might not be happy with Mr. Ignatieff forcing an election

    Big come back in Winnipeg North for the NDP and Fantino just hangs on to Vaughn.

  32. I have Coady holding on by a slim margin.

  33. New Harris Decima Poll...Harper lead drops to 8 points, Ontario statistical tie between Liberals and Conservatives.

    The latest poll puts Tory support at 36 per cent, Liberals at 28 and the NDP at 15.

    Back to the drawing board.

  34. any chance your new model can be applied to Ontario provincials? i know it'd involve a whole new set of coefficients for stars etc, but at least the ridings are the same :P

  35. Bryan,

    I'll be setting up similar models for every province as their elections get closer.

  36. Why is it the numbers for the ridings change form each time I look at it...Miramichi earlier had the conservatives in front by about 9%???

  37. For all the talk about the so-called 'Ford Effect', it's interesting that your model still shuts the CPC out of the 416. 20 Libs, 2 NDP and 0 CPC if I count right.

  38. I think the new model does look pretty accurate, but really I'm just looking at the numbers from the last election and the resulting seats. I can't really say the changes surprise me much.

  39. Peter

    In response to anonymous thought that things change with the new harris decima poll

    It will have negligible impact on your seat prediction (unless there is drastic chance in vote concentration) as the
    CPC 36 Lib 28 NDP 15

    is really close to your amalgamated weighted polls. 39 - 28 - 15.5

    It will take a lot more drastic different polls than the HD poll to change your seat predictions.


  40. Tommies,

    The numbers did not change. It might just be your memory. Those are the same riding projection graphics I put up earlier today.


    I will have to look at the regional results. I imagine some changes, a few seats here and there, but nothing major. The projection doesn't swing wildly with just one poll.

  41. Michael Barkusky20 March, 2013 16:51

    I can't see anyone dislodging Elizabeth May in Saanich Gulf Islands, nor can I see the LPC or Cons defeating Don Davies in Vancouver Kingsway. Both are strong constituency MPs and both are well-aligned with the small-g "green tide" that is running pretty strongly in many parts of BC. This is, I think, ocurring in reaction to the exceptionally "brown" governing agenda of the Cons and the overall weakness (Joyce Murray excepted) on climate change that has characterized the post-Stephane-Dion LPC. Many people out here don't see environmental irresponsibility as a smart "economic" strategy. Ecological economics is simply better economics than Harper-Flaherty natural-resource liquidation economics, and a growing number of voters can see this.

    1. Please note, you are commenting on a post that was written two years ago.


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.