Sunday, May 1, 2011

Final Projection: Conservative minority, NDP runner-up

ELECTION NIGHT UPDATE: Well, that did not go well. I'll do a post-mortem tomorrow and investigate whether it was the seat model or the popular vote model (or both) that was wrong. The seat model will definitely be closer to the real result when the correct regional/provincial vote numbers are put into it, but how much closer will have to wait until tomorrow.

The polls were released later than expected, and so the final projection update is also coming later than expected. Many polls were added to the projection today: 10 in all. They include polls from Nanos Research, Léger Marketing, Forum Research, Abacus Data, COMPAS, EKOS Research, CROP, and Harris-Decima. And several of these firms had multiple polls added to the model as they released new data throughout the day.

In the end, this amount of polls has represented several normal days' worth of movement in the projection. The change has been radical.

On the eve of the vote, projects that the Conservative Party will win another minority government, with the New Democrats forming the Official Opposition. The Liberals will place third while the Bloc Québécois will be relegated to the fourth position in the House of Commons.

The Conservatives are projected to take 36.4% of the vote and 143 seats, the same number of seats they had when the government fell. However, considering that two of the three vacancies were safe Conservative seats, we can even say that this is a loss of two for the party.

The New Democrats are projected to take 27.3% of the vote and win 78 seats, an increase of 42 over their pre-election standing. It is almost double their previous best under Ed Broadbent.

The Liberals are projected to win only 22.8% of the vote and elect 60 MPs, a reduction of 17 since the election was called.

The Bloc Québécois is projected to win 28.1% of the vote in Quebec and 6.7% nationally, enough to give them 27 seats in the province. That is a loss of 20 for Gilles Duceppe, and the first time the Bloc would be reduced to a minority of seats in Quebec.

The Greens are projected to win 5.6% of the vote and no seats. No independents are projected to be elected.
Compared to yesterday's projection, this is a drop of one seat and 0.4% for the Conservatives and a drop of 1.3% and five seats for the Liberals. The New Democrats are up an astounding 19 seats, as well as 2.2%. Most of this comes in Quebec, where the Bloc drops 1.7 points and 13 seats.
Twenty-two seats have changed hands in the projection since yesterday.

In British Columbia, the NDP takes Esquimalt - Juan de Fuca from the Conservatives, who had taken the seat from the Liberals in the projection recently. The New Democrats also take Saskatoon - Rosetown - Biggar from the Tories.

In Ontario, the Conservatives take Ajax - Pickering, Brampton West, and Guelph from the Liberals. The New Democrats take Beaches - East York and Parkdale - High Park from the Liberals and Oshawa from the Tories.

In Quebec, the New Democrats take Charlesbourg - Haute-Saint-Charles from the Conservatives. They also take the following seats from the Bloc:

Alfred-Pellan, Brome - Missisquoi, Compton - Stanstead, Longueuil - Pierre-Boucher, Marc-Aurèle-Fortin, Québec, Repentigny, Rivière-des-Milles-Iles, Rivière-du-Nord, Saint-Bruno - Saint-Hubert, Saint-Hyacinthe - Bagot, Shefford, and Verchères - Les Patriotes.

This has been a very exciting ending to a very surprising campaign. I am confident in my projection, but I am well prepared for it to be wrong. This campaign has proven that anything can happen, and that Canadian politics do not always have to be boring.

Tomorrow night will be a very exciting one, no matter what the outcome. I will be watching simply as a Canadian, but will be on Twitter to comment on how the night unfolds and how the projection holds up.

And now that I've provided my take on what will happen tomorrow night, it is up to all of you to go out there and prove me wrong. That means you need to vote!


  1. Thanks for sll of the hard work. Hieronymus

  2. But what will the "Bin Laden" effect be on electoral turnout and last minute choices?

    As if things weren't unpredictable enough already!


  3. This comment has been removed by the author.

  4. Any comments on how tonight's big news could affect tomorrow's election?

  5. Thanks for all the hard work over the election campaign Eric! Now we all wait and see.

  6. Sounds reasonable. Mine are that far off either. As for the Bin Ladin effect tough to know, although I suspect a lot in the Tory camp are probably happy about this of any group.

  7. What's the twitter of ThreeHundredEight??????

  8. Beat me to it anonymous.... I suspect myself that this could cause some of those voter of of fear (ie conservatives) to switch votes.

    Seeing as there was the poll that asked about strategic voting and the large numbers that would seem to sway things is there anyway you could release a projection with that scenario?

  9. While I admire your well thought out assessment, I just have a strange sense that the NDP surge is going to be stronger than any of us expect. I don't know why I think that, but it's just a feeling in my gut.

  10. First off, great site, I've been following for weeks.

    Can you put a note on how many seats your projection has within 1 or 2%? A quick glance shows that the NDP are within 1% or so of another 6 Bloc seats in Quebec - including Gilles Duceppe's riding. My guess is that even if there were no new polls tomorrow, the simple fact of aging the older polls would flip some of those seats.

    Thanks for all your work on the site,

  11. Eric;

    Been reading your blog day in and day out throughout this election and cannot thank you enough. I tend to agree more with your calculations than the pollsters (i.e. NDP with 100+ seats). But this election is so fluid anything can happen.

    I'm looking forward to seeing how the results tomorrow compare to your numbers! The death of Bin Laden may throw a wrench in favour of the conservatives (war on terror) but I have a feeling it is too last second to be of substance.

    You mentioned you were going to make a separate analysis based solely on last weeks results. Will you still be doing this? Would be interesting to compare these results and last weeks result to see the differences.

    Thank you again for everything you have done. Will of course be voting tomorrow.


  12. same "number" of seats (not 'amount').

  13. I doubt if the death of OBL will make any difference - but if it does have an impact it will put people in a joyful, euphoric mood and that will make them more likely to vote for the party and leader who is optimistic, forward-looking and hopeful (guess who).

    Harper could have benefitted from a terrorist attack on Canada since he's all about paranoia and darkness and pessimism. No suh luck Harpo.

  14. As an American who just started following Canadian politics a week ago, what are some websites/twitter accounts people would recommend to follow the election results in detail tomorrow night?

  15. In the end, the latest polls did not differ too much from my regional findings, so doing a projection for only the most recent polls isn't really necessary. Plus I'm very tired.

    The only main difference is in Quebec, and you could add another 10 seats or so to the NDP, from the Bloc. But I think my numbers will be closer to the mark than the 40%+ we've seen in Quebec.

  16. The death of Bin Laden would probably means less support for pro-war... But not big turnover.

    Big work here for this prediction. More realistic then those in the past days. One mistake form my perspective, NDP will win in Jonquière-Alma.

  17. I doubt there will be any Bin Laden bump. As far as we know now, there was no Canadian involvement. If this story had brok a week ago, there might have been some effect if one of the leaders or an operative had said something foolish.

    Now its too late.

    Remembering back to the Charlottetown accord, the Blue Jays won the World Series two days before. Although the pro-accord people had used pictures of the Blue Jays winning in their TV ads, a neutral good news story did not change how we thought of that vote.

    Canadians take this too seriously to be swayed by something not about Canada.

  18. Thank you very much for all the hard work, Eric. Thanks especially for working so late tonight to produce this last update.

    You have built quite a following. Congratulations.

  19. The death of ObL will likely help the Conservatives.

  20. Thanks for the final predictions. I suspect they will be close to accurate except there is one independent in Qc that is likely to get re-elected (Portneuf-Jacques Cartier). Time will tell...

    QC anglo

  21. a different Eric02 May, 2011 00:11

    Thanks for all the hard work Éric, I've really enjoyed your analysis throughout the campaign. Would you consider doing a projection with only the last 5 or 6 days of polling results? I'd be very curious to see what effect that would have, especially on the BQ/NDP numbers in Quebec.

  22. You're going to be low on the Tories for 2 reasons.

    1) The Big Blue machine's ability to get out the vote - particularly in the face of the poor Liberal fortunes and the volatility of the NDP vote will mean they'll get 5 more seats than you think they will.

    2) Polling traditionally lowballs the Tories, highballs the Liberals and just plain screws up on the NDP. Given that your model is entirely built on polls, you will experience the same problems they do. For instance, there's no way Peter MacKay loses in Central Nova, despite what the polls are telling you.

    Next time, you might want to add in a personal veto on certain results (unless you're looking for a purely poll-based analysis).

  23. Will be fascinating to see if you are correct about Ajax-Pickering. The Tory candidate there has seemed an awful lot weaker than expected - very flat campaigner, not an even match for the Liberal. The assumption that the NDP will significantly improve here over the 2008 result also seems generous given that their candidate famously went on vacation for the first half of the campaign.
    - Gilbert

  24. Given the combined impact of vote-splitting and the impact of limited resources on their get-out-the-vote efforts I tend to agree with your analysis of the scale of the NDP surge. Tomorrow night will certainly make for interesting watching!

  25. I hope you're wrong and the liberals keep enough seats to be the offcial opposition....thanks for your efforts, cheers!

  26. Gotta call it like i see it: What a pant load of a prediction. Try this:

    CPC - 155-160
    libs - <60
    NDP - <60 (neck and neck with the liberals)
    Bloc - <35
    and 1 independent

  27. Eric Said:

    "In the end, the latest polls did not differ too much from my regional findings, so doing a projection for only the most recent polls isn't really necessary. Plus I'm very tired.

    The only main difference is in Quebec, and you could add another 10 seats or so to the NDP, from the Bloc. But I think my numbers will be closer to the mark than the 40%+ we've seen in Quebec."

    I kind of had a hunch it would result in something more or less similar as your projections have already moved dramatically the last couple days, today the most of all due to the sheer amount of polls delivered.

    If your predictions are close than no need to check. If they happen to be way off, than perhaps worth creating the new projection after the fact to compare.

    I'm guessing tomorrow there will be little in the way of updates on this site as we will all be watching elections. Will you have your post-election commentary on Tuesday?

    Unfortunatly do not have Twitter. Does 308 have a Facebook account?

  28. It's gonna be a wild ride when the results start coming in.... especially Quebec to see how far the NDP rises and the BQ falls, and Ontario to see if the (seemingly slim) chance of a CPC majority stays alive and to see how far the LPC falls.

  29. @Ron

    Try www, non partisan, good analysis. online streaming too.


  30. Website;
    Twitter account: acoyne and others in his timeline

  31. I don't think Osama B. Laden death will have any influence on the final results. More like wishful thinking from those who think it will affect CPC negatively or positively.

    I think the forecasted rainy day/bad weather in the GTA area will have more of an influence. Same for the advance voting.

    Also, the majority/minority question for the Cons will depend more on the final number for the Liberal rather than final number for the NDP. If the Libs go under 20%, then NDP will need a serious support over 30% in order to stop a Harper majority.

    Last election all predictions were at least 8 seats under the CPC final count. Interesting to see if the crystal bowl capabilities of political junkies have been improved after 2.5 years.

    All current leaders are past their expiration date and I think country, as a whole, will benefit for a change of guard. And, I'm willing to bet that my wish will be mostly fulfilled in the next couple of months.

  32. 308 has no Facebook account, but you don't need to "have" Twitter to be able to use it.

    Just click on the Twitter image in the right-hand column and you can read my tweets there.

  33. Amateur Psephologist02 May, 2011 00:28


    Bravo for all of this. You win the prize - no matter the outcome - for all the invention, dedication and brainpower that it takes to create this wonderful tool for the rest of us to use.

    Thank you.

    And if your end of the range is closer than mine, I will be among the first to congratulate you.

  34. Thanks for all the work on the site! I've been following your projections and commentary for weeks. It's always been a good read.

    While I think your model reacts too slowly to accurately reflect what will happen tomorrow, I have appreciated the level methodology you've applied along the way.

    Please vote!

  35. Had OBL been killed in Afghanistan, this might have given the Tories a boost. But since he was killed in Pakistan, it may make some people wonder why we're in Afghanistan.

  36. Thinking the Bloc might go down a bit more, but probably wont lose official party status.

    CPC 137
    NDP 98
    LIB 56
    BQ 14
    GRN 1
    IND 2 Guergis, Ford

    Actually have the Conservatives gaining 6-8 in Ontario in this scenario, counterbalanced by losses in Quebec and BC specifically and a couple others across the country. Maybe a bit of wishful thinkng with the BC result, but i think most polling has it at a virtual tie and that likely gives NDP 14 seats as well (2008+ EJDF, Newton North Delta, Surrey North, Vancouver Island North, Kamloops,) as well as the Greens electing May.

  37. Of all the seat prediction websites this is my favourite. Thanks for all the hard work!

  38. I suspect that there will be more NDP and CPC seats at the expense of the LPC and BQ. But I don't need to see any other projections, because I don't see any change in the balance of power without some significant divergence from polling or some oddly fortuitous split which though significantly possible; polling wouldn't be able to see anyway.

    Conservatives in government, and the NDP's only shot is if they can rope in the Liberals AND get the Bloc to abstain. (I can't see that happening) I get the feeling there'll be more abstaining from the LPC since the NDP and Bloc will see little reason to do so and the LPC will be changing leaders. (Why they can't learn to vote down motions and abstain on confidence motions I don't know, they'll have to figure out if they can do that soon)

    Is this opposition stronger with new members and a clearer ideology, or will it be hamstrung by same as was the ADQ? Will someone try to pick at the remains on the Liberal vote, or will they ignore it? We'll see.

    Good job Eric.

  39. "Last election all predictions were at least 8 seats under the CPC final count."

    ...that was the case in 2008, but in 2004 and 2006 almost everyone OVER-estimated how many seats the Tories would get and they under-performed on election day - so who knows!

  40. BQ projections are too high. NDP will take atleast 3-4 more away than the prediction. Wouldn't be at all surprised to see BQ with less than 20.

  41. Your Quebec projections are going to be terribly off.

    First of all, your popular vote in Quebec (34.2%) prediction is completely at odds with recent polls which have the NDP above 40% and as high as 45%. Your method gives too much weight to old data and you're not picking up the extent of the change of voting intentions in Quebec.

    For example, (just going aphabetically) take a look at Abitibi Baie James. You predict a Bloc win but it seems to be clearly going NDP. I am cross referencing with Project Democracy. They actually seem to predict a Bloc win too for some reason but if you look at nanos, epkos and decima, they all predict and NDP win since after April 25th.

    Find me on Twitter: pjanuk (It's not a very political account so don't be thrown off).

  42. Thanks Eric;

    Forgot to write my psedonym last time with the Twitter question. It's Non-Partisan. Will check on Tweets tomorrow.

    As a comparison to another site I respect for projecting polls: Canadian Election Opinion Polls (CEP):
    Conservatives: 143 (Eric) and 146 (CEP)
    Liberals: 60 (Eric) and 65 (CEP)
    NDP: 78 (Eric) and 61 (CEP)
    BQ: 27 (Eric) and 33 (CEP)

    Witht he exception of NDP the differences are not overtly large. This compared to the polls predictions of 100+ NDP shows that using a methodological approach is a) far more conservative (not the party) and b) far more realistic.

    The polls seem to want to grab attention with their numbers while both 308 and CEP attempt to gauge the election based on a riding by riding estimate.

    It will be a true test whether poll estimates or riding estimates is the best way to determine actual election results.


  43. The big news in my eyes is your projection now shows the combined NDP, Grit popular vote as being 50.1%.

    Heh, clear "mandate" from the people if I've ever seen one.

  44. Eric, Thanks so much for an enormous amount of work. It's still anybody's ball game. We do hear about party GOTV capabilities and Liberal voters staying home in 2008. My feeling this year is that the voters who in previous elections stayed home will be getting themselves out to vote this time around. Once Layton started showing some traction, i.e. increasing poll numbers in Quebec, he became somebody that the anti-Harper majority could rally around. But until the Ontario results come in tomorrow night, a Harper majority remains an outside possibility.

  45. @Ron, I would recommend following the CBC's online election coverage. The pros really are better at covering things like this, and it's more fun to watch it live than to refresh internet pages.

    If you can't stream video, you can try the major national newspapers: "Globe & Mail", "National Post", "Toronto Star". And maybe some smaller ones like "The Tyee".

  46. Ok I did what I asked earlier including strategic voting into the poll. I gave 33% of Liberal vote to NDP candidates with the best chance of beating bloc or cpc frontrunners.

    With NDP I put 22% swinging to Liberal in contested ridings. I also gave May a few swing votes to defeat the CPC candidate.

    My final seat projection based on strategic voting is:

    CPC 133
    NDP 86
    Lib 67
    Bloc 21
    Grn 1

    This puts a Lib/NDP coalition within 2 votes of majority without the Bloc

  47. this is so wrong. we are going to see a 160+ seat tory majority!

    the conservatives show up more on election day than teh liberals and NDP, so expect 43% vote conservative

  48. Beau travail Éric! Ce fut un plaisir de suivre l'évolution de tes prédictions tout au long de la campagne électorale. Reste à voir si le vote NPD sortira aussi fort que prévu au Québec.

    Cette montée d'un "tiers" parti au Québec n'est pas sans rappeler la belle performance réalisée par l'ADQ en 2007 (élection provinciale québécoise): ce parti était alors sorti de la marge pour devenir l'opposition officielle. Les sondages les plus optimistes n'avaient pas été en mesure de prédire.

    Sylvain Bérubé, Sherbrooke

    (if needed: Google Translate)

  49. Ron - Great that you're interested. There are laws banning tweeting and passing on the results before 8pm PST, because we have 6 time zones across Canada. No cheating on the west coast.,, just a couple of the mainstream media outlets that will be covering it.

  50. About the whole the Conservatives show up more arguement: has anybody seen proof of this beyond elections where the conservatives are assumed to be succeeding at their goals?

    I am somewhat dubious of a theory of Conservatives being more likely to vote if the data for that is only 2006 or 2008 federally.

    I grant you that there is intuitive evidence based on polling about intentions. But is there hard fact to back up this idea?

    There is some significant doubt that the CPC will get what they asked for - a majority. Although we may not have an exact same situation as this, don't we have situations provincially where the conservatives were obviously slatted to lose? Did the conservative vote strength hold up in those situations? (1993 would not really count in this regard, BTW, given the conservative vote split that time around.)

    Cause, without some data, it sounds like spin.

  51. "...that was the case in 2008, but in 2004 and 2006 almost everyone OVER-estimated how many seats the Tories would get and they under-performed on election day - so who knows!"

    :) no need for upper case.
    Exactly, the seat projections tends to overestimate the opposition and underestimate the party in power.
    Happened, in various degrees, in each of the past three elections and, chances are, it will happen again. However, if tomorrow we see long lines at the voting stations, Cons might be toast.
    Atlantic Canada and W Quebec vote will give an early indication on how things will evolve (mainly how Liberal and BQ vote holds). Hopefully, we can get some results twitted as early as possible.

    Social-media and strategic voting web-sites impact are all BS. Nothing more that static noise in the progressive echo-chamber. Don't expect to influence the final result one tenth of a percentage.

    Interesting election for sure though (by Canadian standards).

  52. The polls close in BC at 7pm actually

  53. "The big news in my eyes is your projection now shows the combined NDP, Grit popular vote as being 50.1%."

    I don't see why people automatcially assume that the Liberals and the NDP are essentially interchangeable. The combined Grit and Tory vote is also over 50%, and those two parties have much closer platforms than either does with the NDP's. I know that a lot of lefties have a ABC attitude, but really, they should remember that the Liberals have historically been willing to ally with the NDP if necessary because they view the Conservatives as their real opposition -- i..e for opportunistic rather than ideological reasons.

  54. Voting Day IS Remembrance Day!

    If you don't vote today,
    then don't be a hypocrite...
    by wearing a poppy on November 11th....

    British parliamentary democracy is the result of the struggle between rulers and ruled for 1,000 years. They gave us the right to change our Kings -or not- peacefully!

  55. For me its looks like:

    CPC 133
    NDP 93
    Lib 65
    Bloc 14
    Grn 1
    Ind 2

    But we'll see if the Orange Wave turns in tsunami. If so we could see an NDP majority government as per the highest possible vote percentage they could get.

    Regardless its going to be an interesting night.

    Thanks for the hard work Eric... still think the polls and pundits don't know how to adjust for what I expect will be a 5% rise in voter turnout.

    A Proud Canadian Partisan

  56. Oh God... in your model, Duceppe wins his riding by 0.2%.

  57. Eric,

    A very big "thank you" - reading your blog has been a highlight of my every day since this election began.

    I really appreciate your analysis and insights. Keep up the good work.

    Philippe Theriault - Southern Ontario

  58. I have revised my earlier assesment based on the values for strategic voting provided by IPSOS REID and manually entering them into excel.

    The results based on best possible scenario of LPC/NDP strategic voting based on the poll are:

    CPC 132
    LPC 66
    NDP 94
    BLOC 15
    GPC 1

    This gives the LPC and NDP enough seats to govern without the bloc.

  59. The fact that Canada still uses plurality voting is a major problem. As Robert mentions above, the Tories still have a chance because the progressive vote NDP, Liberals, can easily end up being split in specific ridings. Of course, if the momentum remains with the NDP the voters may decide to engage in a sort of preferential voting in their own right…

  60. @Joel R

    Sorry, but I'm going to place more faith in the well-explained model this site has provided us than the unsupported ravings of someone who clearly has a partisan axe to grind.

  61. Id love to see an NDP official opposition to a Tory majority. I think Ontario is the key to a Tory majority and I think the vote splitting and Blue Liberals will hold their noses and vote Tory.

    Still sticking by my 3-4 seat Tory majority as I said many weeks back.

    Great work here Eric.

  62. A huge round of applause for Eric and all his great work over the campaign.

    Thank you so much for all you've done here - I've really enjoyed reading daily.

    Well done!

  63. I will also be looking to see if there was any impact from all the strategic voting websites. If there was, it may show up in the narrow Lib/Con races in Ontario like Ajax-Pickering and Guelph, and in the narrow NDP/Con races out west like Kamloops-Thompson-Cariboo, Nanaimo-Alberni, and of course Saanich-Gulf Islands for Elizabeth May.

  64. Cons NDP Lib Bloc

    BC 24/ 10/ 2
    AB 27/ 1/ 0
    SK 12/ 1/ 1
    MB 9/ 4/ 1
    ON 62/ 20/ 24
    QC 11/ 27/ 8/ 29
    NB 7/ 1/ 2
    NS 3/ 4/ 4
    PE 2/ 0/ 2
    NL 2/ 2/ 3
    Terr. 1/ 1/ 1

    Total 160/ 71/ 48/ 29

  65. Great job and thanks for your work.

  66. The interesting thing is the BQ can still lose 11 more seats to the NDP if trends continue to the polling booths.

    But it is interesting to note that most of the Liberal losses occur in very tight races so it should be interesting to see how the night goes!

  67. Not sure about Guelph. The Liberal candidate has strong community ties, versus the CPC who has missed 7 of 9 debates in the community. In the last election, it was a closer race because the CPC candidate was better and tied to the community. I think Frank will hold Guelph. One less for CPC!

  68. Eric,

    A million thanks for all your hard work. I haven't commented but I've been following your blog with intense interest for the past couple of weeks & it has been engaging, stimulating & informative. I really appreciate you taking the time to share your expertise & analyses with us. It's been quite a ride - can't wait to see what actually happens today - but I'm hoping your numbers are just about bang on. Bravo & Merci!

  69. Thanks for all the work, and providing a place for us to come discuss :)

    My best guess:
    161 CPC
    62 NDP
    55 Liberal
    29 Bloc
    1 Independant

  70. Indeed Eric thank you very much for all the work.

    Now out of the trenches and into the battlefield !!

  71. My prediction:

    NDP 127
    CPC 121
    Libs 41
    BQ 18
    Green 1

    No regionals show majority for anyone, but Layton has the big mo going in and Canadians are really angry at Harper. The last minute smaer on Jack is sickening even some Conservative voters.

  72. I love how the CPC supporters are predicting majority when the evidence is massively against it. LOL

  73. Thanks for you hard work. I really hope May can pull it out in her riding. I've heard that Green Party polling can often underestimate their support because some polling companies just ask if you are voting CPC, LPC, or NDP and don't present GPC as a prompted option. Is this true and do you account for it in your model?

    Thanks again!

  74. good job Eric,
    I have been a regular visitor to your blog for almost 18 months, (so long now?) since I started logging on ploitics, and it is definitely a useful and engaging tool. Congratulations on parlaying it into what appears to be a succesful little enterprise.
    I look forward with interest to comparing it with actuals tomorrow.
    btw, you should incorporate a forum now that you have a decent amount of traffic, it will engage your visitors more deeply, and give people a cross-party forum just to discuss politics, that isn't associated with any particular Party brand.

  75. Thank you Eric for all your hard work. I've been addicted to this site for weeks. I will miss checking in everyday now that the election will be over.
    a grateful Manitoban

  76. Eric:

    I cannot thank you enough. Great job and a great service to the Canadian public of all political stripes.


  77. Hi Eric,

    Do you by any chance know the result for the Bloc nationally in the last Forum poll? The Hill Times article doesn't say.


    Christine D.

  78. The big bump in the polls and the tougher more critical look that people were taking at Layton changed the way he was campaigning.

    He turned down a interview request on Question Period!! That has basically been his infomercial. Layton walked away from reporters scrums.

    Harper had a strong interview with Craig Oliver. Harper never is on QP.

    This is sort of like 2004 where Martin went all out campaigning and Harper stopped. There was a huge Con-Liberal shift that the polls didn't catch.

  79. An Easterner in Northern Quebec02 May, 2011 10:42

    I've been obsessively following varioud polls, national and regional... and this site is definitely one of my favoured resources, both entertaining and informative, and I actually expect to be quite accurate.

    I've taken into account a number of ridings in which the Conservatives are running close and given it to them in a "worst case scenario" fashion... as well disregarding some polls I know are wrong such as Esquimalt-JdF and Central Nova... and came up with this:

    In alphabetical order:
    BQ 27
    CON 152
    LIB 54
    NDP 75

    So I'm thankful... My first child is due in december, and I can't afford private healthcare. I would wager a large number of us in Canada can't, or at the very least the higher-than-our-proportion-of-federal-deductions premiums the private companies would charge - talk about impacting our take home pay! I digress, I'm just overprotective of my child to be and his/her future... Ignore my rant :)

  80. Great work, terrific information! Thanks very much. Also thanks to the commenters for keeping it about the results / process vs. partisan snipes. Very refreshing vs. what I read elsewhere. Much appreciated all around. dh

  81. Fun election. From the federal battle become NDP-Conservative to my local riding having a no-show NDP (I'm a Green) mixed with the federal Liberal collapse perhaps leading to the Greens moving into 2nd locally. I expect an entertaining night of TV viewing.

    Part of me keeps wondering if the NDP could do the impossible and get an outright victory and if Harper's head would explode should that happen. Just imagining the various Sun headlines on Tuesday should the NDP pull it off (picture nuclear bombs going off, end of Canada as we know it, etc).

    What if the Liberals pull a PC 1993 and end up in single digits for seats - could it happen?

    Potentially we could see 3 federal parties nearly collapse if the Liberals drop down as badly as they look to be, if the BQ keeps going down, and if the Greens fail to win a seat and get sub-4% support. Or we could see a 5 party HOC with a minority lead by the NDP.

    As to seat counts - what would I guess? I have no idea really. I figure Ms. May will win her seat, the NDP will crack 70 at least (could they get 100?), the Liberals drop below 50, the Bloc down to sub 10, leaving a majority for the Conservatives unless the NDP does 100+ in which case Harper, Iggy, and the BQ go away forevermore (thank goodness).

  82. Eric,

    Your model seems to still have a significant time lag. I don't think at this point, right before the election day, when there's been a huge dump of data (>20,000 people polled from 28 April-May 1), time lag is justified at all. I believe the correct step would be to only use the data from say last 4 days - and that would be your maximum time lag. I analysed the data from all 9 polls for these dates (N=20608 polled), and computed average popular support for the parties.

    Block: 5.6, CPC: 35.9, Green: 5.6, Lib: 19.7, NDP: 31.7

    You can see, that your number are suffering from the time lag, as NDP is much lower in your prediction. Based on the last three days only, the difference between CPC and NDP is only 4.0% (yours: 9.1%). This is way too large to be explained by factors like pollster bias (they almost all seem to be agreeing in the last four day), so it's definitely the lag of your model.

  83. Maybe there are already too many comments here, but me too, I really liked this site.

  84. Great site!
    This election is likely the most difficult one to call in recent memory. My gut tells me that the CPC will do no worse than your model and given the commitment level of its voters will likely do better. The big wildcard is how firm the NDP vote is. Given the recent upsurge you could argue two ways. One, the vote is soft and many voters may back away and return to one of the two traditional parties. Or two, the party has momentum that will move them even higher on election day. I'm predicting the former. Either way, this is one election that will be studied extensively by political junkies.


  85. Harper is going down for his dishonesty and smear campaigns. The Surge of the NDP is a direct result of Harper's failure and nasty ways. People are angry and that is why the NDP is rising and the turnout is high.

  86. Thanks for your good work.
    As a native Quebecer I've waited for this day all my life. Wherever he is, Robert Cliche must be wearing a big smile today!

  87. Éric, thanks for all your work. I went to this site daily during the entire campaign... well, more than daily actually.

    However, I have to say that, even though I like this site, I think it is part of a trend in the media that is detrimental to politics. This trend is about covering the "horse race" instead of the campaign itself. In other words, people are talking of the movement in the polls and the possible results of the composition of Parliament and they almost don't say a word about the real issues of the campaign. Political analysts have become too big for their breeches, too "meta", or simply too popular, so that we hear only about "strategy", "targeting", "momentum", etc...

    What about health care? The economy? The respect of democratic institutions? Don't care, doesn't matter, it's all about the race.

    Who's in first? Who's in second? Who is moving ahead? Who is moving back? Now that reels the ratings in!

    Yeah, it's exciting, but the campaign is about the future of the country. It will place politicians in power who will shape the trajectory we are on and affect our lives differently (no matter what cynics say). We talk a lot about a possible Conservative majority of a lib/NDP coalition, but we don't talk about what it would mean in terms of policy, what it would change for Canadians from coast to coast.

    Which is why, even if my "horse" is doing extremely well this time around, I feel disappointed by the campaign.

  88. Simon,

    I don't disagree, but it is not my job to report on the policies of the parties and the issues facing the country. I am not a journalist.

    Polls and public opinion are part of politics and help inform voters. Knowing that 30% of Canadians are saying they will vote NDP instead of the usual 15% or so is important information. You might think it is better if we all voted in a vacuum, but democracy is more than just the sum of individual choices.

    And I think you might be barking up the wrong tree in blaming the media. If Canadians wanted reasoned, in-depth discussion of policy and issues, newspapers and television would act to meet that demand.

    You might say the media has an obligation to provide this information, but that is another issue. Even the parties try to provide this information, but a lot of people are more interested in personality.

    It's a cycle, and we are all a part of it. I like in-depth discussions of policies and issues, but I'm not equipped to provide them. Instead, I provide projections and poll analysis, which I also find interesting. A lot of Canadians seem to agree, and I don't think they should be judged for it.

  89. Old polls can represent two things
    -advance voters (2 million people voted in advance)
    -voter turnout organization

    You could try a model that uses the % of advanced voters in each riding (Election Canada published this) to average the poll from a week ago with those that are current.

    I'm not sure how you could include organizational strength.

  90. It's apparent that know one really knows how many seats the Conservatives will win. Everyone is certain that the NDP will be the opposition and this election has been a humbling experience for the Liberals and Bloc.
    I just can see this NDP/Liberal coalition at all. The Liberals are too proud to accept Layton as their new Leader.
    Excitement is what we may be in for,over the course of the week. Our Governor General may have a few visitors in the next few days

  91. My pretiction :

    CPC 131
    NDP 88
    LIB 65
    BQ 21
    GRN 1
    IND 2 Guergis, Ford

  92. Pardon me for not what can be added

  93. Éric, don't misunderstand me. I don't think people should necessarily vote in a vacuum, without knowing where the parties are at in polls, especially considering our FPTP system where people more often vote against other parties than for a party. Since all the votes that go towards losing candidates are basically "lost", they result in zero representation for that party, voting for someone over the other basically cancels all the votes that other guy received. This is unlike proportional systems where every vote counts towards electing the party you're voting for. When parties ask for support to "block" another party, what they're basically saying is "Vote for us so as to cancel the votes of the voters who vote for that other party".

    As to answering to the demand... I think this presupposes unjustly that the public opinion is fixed, static, that it cannot be modified. But companies put billions of dollars into marketing betting that it's not true, that the public can be influenced.

    I think the media should do this, that there is a part of an obligation to try to bring the level of discussion up, not down. I even think this would make elections more popular, think about the referendums we had. It seems to me people always go out in huge numbers when the issues dominate instead of just a horse race.

    I think however that there is another reason why the media prefers to cover the race and not the issues: the desire for objectivity. It's hard to be objective when you cover the issues, because you have to analyze the offering of the different parties and it's quite hard to avoid the impression of bias. If a politician keeps lying or saying misleading things and you keep correcting what he is saying, then you risk being attacked as being against him and for the other guy. The media is sensitive to these accusations of bias and try to tone down the coverage to avoid having to "take a side", covering instead the race.

    The parties do try to diffuse that information, but they are limited to what the media then transfers to the public. Huge policy speeches are ignored, or summed up as one sentence, then they show a clip of a party leader answering a journalist question about his party's standing in the polls.

    I do think that it is possible covering the issues could be more involving, not entertaining, because entertainment is something we look at and enjoy, and debating the issues involves people and their opinions.

  94. Thank you so much for this blog and all this hard work!

  95. Excellent work, thank you. I just wish you had also calculated the ceilings with the last batch of polling data.

    Maybe something you could add in the future would be error ranges and confidence levels,and so forth... For example, you could say "within a 70% level of confidence, I project that the CPC will win between 142 to 144 seats... and within a 90% level of confidence, I project that the CPC will win between 139 and 147 seats." You get the idea. Anyway, awesome site, thanks again.

  96. Your results have been fasinating - I eagerly anticipate the results tonight to see how good statistics can predict an election like this!!


  97. Neil in Ottawa02 May, 2011 15:40

    Eric, I also want to say thank you for this.

    Also, don't worry if your projections are not as accurate as Nate Silver's. The US system lends itself more to this kind of projection, as each state has so many polls and large sample/population sizes, whereas in Canada we don't really do many good riding polls. Nate Silver also has a massive amount of demographic data to work with.

  98. All I can add is that 2 days ago I was a committed liberal and today I voted NDP. I wonder how many more (Libs and COns) are like me?

  99. Eric,

    To the Anonymous 15:19 point: as I suggested a few days ago,even in the most complicated models, the error propagation can always be computed via a Monte Carlo simulation technique. I've done this before, so let me know if you are interested in the details. In doing so, you can get the proper confidence intervals for all of your predictions - even at individual ridings level.


  100. Ma prédiction

    CPC 149
    NDP 91
    Lib 51
    Bloc 17
    Everyone else 0

  101. This has been fun.
    I'm as excited as a child on Christams Eve!

    See you all in another two years!!! haha

  102. OK I will throw this out there. I maintained a similar database based on previous polls and regional polls. I find the Ekos poll results have shown a consistent progressive trend line in voter movement and intentions to vote.

    I made this seat count projection on Saturday morning after the NDP peaked on Friday and then slipped back 3 points Saturday AM and it factors what so many people have been missing in other seat projections.

    This is a moving target.

    There is an antipathy against the NDP forming a government in Ontario that creates a ceiling for them.

    There are Liberal votes going right as well as left - and some undecideds who are anti-Harper just guessing where they should put their vote.

    I think the point lead in Ontario is significant - the Conservatives have always led and in some polls by 14 points in that province over 2nd.

    The Liberals on the other hand are nationally now at 20% or below - meaning they just get by IF THEIR VOTERS show up and only if they dont bail on them at the box after all that.

    Put that 20% number into perspective: Usually the NDP has a vote in that 20% range and they get it so spread out that it doesnt produce seats. How many times have the NDP had 25 or fewer seats in parliament because they were in that percentage of popular vote? This time the Liberals look to be at risk.

    The Conservative vote in Ontario is very efficient and I think this factor + plus the Bloc fallout consuming NDP poll numbers will produce a result like this:

    CON - 155
    NDP - 101
    LIB - 29
    BQ - 20
    IND - 1

    The Liberal fallout is unprecedented for them. They will not retain 50 seats as your projections suggest. The CPC will maintain just 5 seats in Quebec mostly cabinet members. Liberals will have no seats west of southern Ontario.

    Who is motivated to vote is most important.

    It appears that NDP and CPC voters are motivated to vote on a chance they might win but there is a lack of motivation for Liberal or old Bloc supporters because they know they cant win.

    Two parties give voters something to vote for but the other parties are simply ideology because neither really have a chance to go up in seats and everyone knows it.

    In pockets of the country there is NDP momentum and in other places some motivated conservative voters. Those intentions to actually vote for a purpose (ie, vote CPC to keep the NDP out, or vote NDP for Jack's vision) will drive the results tonight.

    Hard to expect 30 or 50 seats from parties which do not invigorate even their base electorate to vote for them. If anything Bloc and Liberal campaigns are leading voters to select one of the other two contending parties.


  103. Just an observer here, but is a centre=left coalition or minority government with tacit support from the Bloc an option? When I was around in the 1990s it seemed the Bloc would clearly favour the left over the conservatives. Or does the Bloc's continued position on an independence referendum put them off limits politically? Or is it by choice the Bloc refuses to support such a governmnet? It seems a shame that with over 50% voting consistently for the centre & left, and less than 40% for the conservatives, that Canada has had so many years of conservative government.

  104. The Conservative numbers don't point to a majority in Ontario. According to the Tories themselves, they need at least 70 seats in Ontario to offset losses elsewhere in the country which is clearly not happening today.

    The Conservatives won't magically sweep Ontario (85-100 seats) like the Chrétien Liberals did in the 1990s because unlike the past Liberals, the Conservatives do not have Ontario numbers anywhere near the 50% range (and the opponents are nowhere near the mid-teens).

    Nor will the Conservatives take at least 70 seats since they don't even have 44-45% in majority of the polls, which are pinning them at the 34-39% range. The most the Conservatives could expect is 60 seats in Ontario since both the Liberals and NDP have their support concentrated in certain areas of the province and not widespread around the entire province.

    Therefore I believe the only party that will emerge with a net gain is the NDP (and maybe the Greens too).

  105. Félicitations Éric pour ton travail!

    Quelle belle élection, bien que peu de viande autour de l'os sur les sujets importants, c'était ma plus belle campagne jusqu'à maintenant! :)

    Pour ce qui est du "feeling" du jour du vote, l'histoire de la vague de l'ADQ a beau avoir été différente, je l'ai vécue, et je n'ai absolument pas de craintes quant au niveau que celle du NPD pourrait atteindre!

    La vague sera forte au Québec je crois... Très forte!

    Bonne soirée à tous!

    Merci encore Eric!

  106. Hello/Bonjour from the UK,

    Here's some target lists for the election:

    I’ve updated the NDP target list to include seats in Quebec vulnerable to a swing of up to 20% since that’s what the polls are showing could happen:

    Here’s the Conservative target list:

  107. Merci pour toutes vos analyses.
    Je commence ma période de sevrage demain.
    Elle s'annonce difficile.
    Merci encore.

  108. Merci Éric pour ton excellent travail,

    J'avais pris l'habitude de regarder ton site, même avant mes courriels le matin.

    J'ai hâte de voir les résultats dans quelques heures...

  109. Great work overall - but turns out you were way off - as all the older polls were. Momentum carried much farther than your back dating data suggested - or even what current polls suggested.

  110. great site, despite the issues with the results. take them into account, weigh the polling agencies accordingly, and come back stronger in 2015.
    thanks again.

  111. I wish I found out about this site before tonight. Great work, it seems like all of the polls were pretty off, everyone knew the Libs would fall off, and the Bloc would follow them, but no one knew quite the extent to which it would happen. Great effort, I definitely look forward to checking this place out in 2015!

  112. Wilson Harron03 May, 2011 00:24

    Long time reader, first time poster...

    It would be interesting to see you go back to your methodology for this election result. There have been a lot of comments about your weighting of previous polls. I don't think you had a bad strategy per se, but some improvements perhaps: instead of a constant decay of old poll results perhaps one that is proportional to the number of days left in the campaign. As we have seen tonight, the closer the polls are to election day, the more likely it is to tell what is going to happen. Maybe you did this, but I couldn't glean that from your description of your methodology.

    On the other hand... who predicted ~40% for the conservatives as the final vote tally? (very, very few pollsters saw this one correctly). Your model did very well, but as they say in computing - garbage in, garbage out - meaning a good algorithm is only as strong as the data that feeds it.

    I wonder how much the voter suppression campaign over the weekend by (?) had an effect on the results.

    Great work on this campaign. I am looking forward to more of your posts - very informative and very well done.

  113. "Nor will the Conservatives take at least 70 seats"

    Boy that was wrong.

  114. Andy JS -

    Thanks for your target list. Any seat could have been a target in Québec for the NDP as results showed.

    I am hangover. My worst nightmare was a Tory majority. I can tell you that everyone in Québec is feeling hangover.


  115. Yeah thanks for all the hard work, too bad it missed the mark by a country mile.

  116. Well, at least you hit the conservative ceiling on the nose. Did anyone anticipate such a complete collapse of the Bloc and the Liberals? Hieronymus

  117. The pollsters didn't see it coming because they only poll a) poll the entire electorate instead of likely voters or b) only polling both decided(more likely to vote) vs. undecided voters (less likely to vote).

    In other words, the model is fine - but the incumbent/government party since '04 has always got a 2-3% poll bump due to incumbency, turnout, or because they didn't account for likely voters (old people).

    So the 37-38% in the polls means 40-41% on election day. Needless to say, I won a lot of beer last night ;)

  118. Let's face it, it was wrong because you used national polling results which have a high margin of error for regions. Also, because of the nature of first-past-the-post, three way splits in the vote create results that you cannot predict from any statistical model. In the past, it has certainly benefited the Liberals, but this time around it definitely benefited the Conservatives (and to a certain extent, the NDP). In addition, a swing of one or two percent of support can added up to far more than one to two percent of the seat total. Hence why the Conservatives got 54% of the seats from just under 40% of the popular vote.

    Really, unless intense regional polling happens, there is no way you can predict how particular ridings will swing. Though there were a few done (like the one that showed May was ahead of Lunn in Saanich-Gulf Islands), it is really not practical. I bet you will find if you look under a microscope, the models needed to reproduce the results will have changed between elections.

    As an example, let's say that 300 people are polled in a survey in Quebec. That averages to be about four people per riding. Though 300 people might be enough to get a general sense of the political atmosphere in Quebec, it certainly cannot be extrapolated to particular ridings, especially to the nearest tenth of a percent in your calculations.

    Finally, you should not be including historical factors into your methodology. If Canadian political history has taught us anything, there is generally a collapse of one party once per decade. This election, it was the Bloc, last decade (arguably) it was the Liberals, 90s it was the Conservatives, and so on. These swings cannot be predicted through polling. For instance, what model could predict that the PC party of 1993 would get 2 seats from 16% of the popular vote?

  119. Well all I'm going to say is I was right. The situation before the polls opened was way too volatile for any polling or other measurement system to have worked at all correctly. Nobody got it right Eric, don't beat yourself up !!

  120. Just for your info, Éric, here's the seats within 1% and their leader and challenger:

    Nipissing-Timiskaming: 0.03% Con-Aspin/Lib-Rota
    Etobicoke Centre: 0.05% Con-Opitz/Lib-Wrzesnewsky
    Dartmouth—Cole Harbour: 0.15% NDP-Chisholm/Liberal-Savage
    Winnipeg North: 0.45% Lib-Lamoureux/NDP-Blaikie
    Esquimalt—Jean de Fuca: 0.64% NDP-Garrison/Con-DeSouza
    Elmwood—Transcona: 0.86% Con-Toet/NDP-Maloway
    Bramalea—Gore-Malton: 0.93% Con-Gosal/NDP-Singh

  121. AverageCanuck03 May, 2011 12:07

    Peter you were wrong to say no majority was possible.

    Eric was very wrong to say no majority was possible for the CPC without Quebec.

  122. I do agree the Liberals have to establish a clear idea of what their relationship is with the left and with the right. Layton very successfully characterized Ignatieff as Harper-light. Harper characterized Ignatieff as out-of-touch. A new Liberal leader would have to be able to overcome both of these legacies.


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.

Note: Only a member of this blog may post a comment.