Friday, April 29, 2011

New Democrats up, Liberals and Bloc down

Three new national polls (not including this morning's EKOS) and a stunning 10 new riding polls have been added to the projection. It has created quite a bit of movement, but still not as much as I'd like to see. Nevertheless, the New Democrats are growing by leaps and bounds and are starting to knock-off Conservative seats in British Columbia and even Quebec.
Nationally, the Conservatives are down 0.2 points to 36.7% but they have gained one seat. They are now projected to win 144 in total. The Liberals are down again by 0.4 points to 25%, while the New Democrats are up 0.9 points to 23.8%. The Liberals have dropped four seats and are now projected to win 70, while the NDP is up six seats to 53.

The Bloc Québécois is at 7.2% nationally and has dropped two seats to 41. That is still a high number considering they are now projected to be at 29.9% support in Quebec. While this is the result of the slow movement of my projection model, I think it will also end up capturing the problem the NDP will face getting the vote out and keeping it there in Quebec.

Note that I will be doing another update here on ThreeHundredEight tomorrow, with my final projection being put up on this site on Sunday.

The Conservatives are wobbling, but they are still firmly in the lead. They've dropped about a third of a point in British Columbia, Alberta, and Quebec, but have gained that much in Atlantic Canada while being stable in the Prairies.

A drop of 0.7 points in Ontario, however, hurts the party. They are now about where they were in 2008, making new seat gains difficult as the Liberals are also about at their 2008 level of support.

In fact, Ontario is the only part of the country where the Liberals are not in free fall. They've dropped big out West, are down another 0.2 points in Quebec, and are close to losing the lead to the Tories in Atlantic Canada.

The Bloc continues to suffer and has dropped below 30% support in Quebec.

The New Democrats, on the other hand, are up 2.2 points in the province to 29.6%. That is their best result in the country. They also gained 0.7 points in Ontario and 0.9 points in British Columbia. They still have some room for growth.

Eight seats have changed hands.

In North Vancouver, the Conservative incumbent Andrew Saxton is now projected to retain his seat. It had been held by the Liberals in the projection for the entire campaign. Also in British Columbia, the New Democrats are now projected to win Surrey North, a Conservative riding. Jasbir Sandhu is the NDP candidate there.

In Atlantic Canada, former Nova Scotia NDP leader Robert Chisholm is now projected to win Dartmouth - Cole Harbour from the Liberals, while former PC cabinet minister John Ottenheimer is ahead again in Random - Burin - St. George's.

And in Quebec, the New Democrats have wrested Portneuf - Jacques-Cartier from independent, Conservative-endorsed MP André Arthur, a state of affairs confirmed by today's CROP poll. The New Democrats have also taken Laval - Les Îles from the Liberals (François Pilon is the NDP candidate) and two seats from the Bloc: Laval (José Nunez-Melo) and Châteauguay - Saint-Constant (Sylvain Chicoine).

107 comments:

  1. It looks like the NDP surge is having at least 2 major effects:

    --> much higher turnout than in 2008 (the advance polls were 35% higher, and the wave is still rising);

    --> higher credibility to the NDP and Jack Layton.

    This Orange Surge is not a statistical anomaly; this is the people of Canada sayin' *CHANGE PLEASE*.

    Honestly: I was very sympathetic to Harper when the campaign began, but now I would vote NDP only for the thrill of it, for change, for fun, to see what will happen, to be also a part of the Orange Wave.

    And then I think: if the wave is affecting even me (a libertarian right-winger), it is affecting a LOT of Canadians.

    The atmosphere will be completely different in Canada after the Election.

    Jack Layton will look like a potential Prime Minister.
    And Harper will look very diminished, nearly devasted. With a target on his back from his own Party.

    I predict the Tories will lose at least 12 seats; and the NDP will have at least 95 seats. Possibly more if tons of Liberals-voters switch to NDP.

    It's a free fall for the Bloc and the Grits; and the Tories are not gaining much from it.

    ReplyDelete
  2. While it doesn't provide riding breakdowns, there's a Halifax Regional Municipality poll out today that points towards Dartmouth Cole Harbour being a Liberal hold.

    http://www.metronews.ca/halifax/local/article/845937--liberals-ndp-neck-and-neck-in-hrm-poll

    Mike Savage's voters are hugely loyal and I'd be shocked to see a carpet bagger like Chisholm steal this away.

    ReplyDelete
  3. Be interesting to see if Tim Harper's projection from the Star holds up ??

    ReplyDelete
  4. It seems that your projections are very conservative compared to the numerous polls that place the NDP at least in several areas, statistically tied for first in others, in solid first place in Quebec. Some are even hinting strongly that the NDP will be the official opposition. How can the NDPs numbers not be translated into more seats? It seems impossible that with polls placing them at 40% in Quebec that they would not get more than 13 seats.

    ReplyDelete
  5. The problem isn't just that your system is slow, this system seems simply unable to capture "game changing" momentum in the last week of the campaign.

    I suppose we'll see how accurate it is in the end.

    ReplyDelete
  6. Is your final projection on Sunday going to be whatever it is that your model says or is it going to be your own personal prediction based off the information from your model?

    ReplyDelete
  7. This model is rapidly becoming out of step with reality.

    It has been days since any poll showed the Liberals ahead of the NDP, but you still do. You have the Bloc ahead in Quebec when polls put them as much as 20 points behind.

    I know the NDP's ground game in Quebec is weak, but that doesn't account for a gap this large. (Nor is the ground-game argument all that convincing; the Tories didn't have strong riding associations in 1958 or 1984 either.)

    I cannot think of any election where there was a gap this big between polls 3 days out and the final results. You are basically claiming that we have the least accurate polls in Canadian history.

    ReplyDelete
  8. Would it be possible not to have a day of "the model is slow/wrong/ridiculous" comments? I'm aware of what my model is showing and how that is different from the most recent polls. I've explained why a dozen times this week.

    ReplyDelete
  9. Seems to me the model works fine. Anyone who reads anything about elections past can tell what popular upswings like this often mean: disappointment and Deputy PM Nick Clegg.

    ReplyDelete
  10. New strength for the NDP is indicated in the Nanos poll today in Ontario and BC, at the expense of both Cons and Libs. This campaign just keeps getting more interesting. If the voter turnout is above 65% it could be really exciting, and if above 70% all bets are off, since that would indicate a swing vote, like the one to Diefenbaker in '57.

    ReplyDelete
  11. Chris in Wakefield29 April, 2011 10:02

    Thanks Eric

    ReplyDelete
  12. I really really hope that Jack wins. Of course I voted Conservative, anyone who holds a shred of common sense would do so. However, this would look wonderful on canadians when we head down the socialist/communist path.
    Remember this people, the rich will continue to get richer and find a way out of the NDP's plans. The poor will get poorer and wonder why it didn't work out the way they wanted it to. Good time coming.

    Luke from Toronto

    ReplyDelete
  13. Seriously, people, it is the same freaking discussion every day. Too much whining that the model moves slowly. We are all aware the model moves slowly - that is part of the point! Eric has already told us that the projection will get a double update on Sunday. Can we all trust that Eric knows what he is doing? After all the work he puts into this site, he is entitled not to be abused in his own comment section.

    ReplyDelete
  14. With respect, if you continue to post updates using a model you admit is flawed, you can't be surprised when people call you on it.

    There's clearly interest in a projection, but I think everyone would appreciate accuracy in the model over continued posts explaining why you're keeping one that doesn't work.

    ReplyDelete
  15. Great site Eric, I'm a long-time reader and I check the site a couple of times each day. I actually think your model will end up doing a good job with the prediction and be a little closer to the result than some of the one-off polls of the last few weeks.

    As for Dartmouth-Cole Harbour and HRM in general, with the results of that poll for the four ridings, I would predict that all four incumbents would win. Stoffer and Leslie are extremely popular so I'm guessing the Liberal vote is fairly strong for Regan and Savage if the Liberals are in the lead (assuming the ridings were weighted fairly equally).

    ReplyDelete
  16. The model is seriously out of step with reality if it projects a Tory win in Random-Burin-St. George's.

    ReplyDelete
  17. @Eric
    I think you're on the right track. You are trying to extrapolate a lot of projections from not all that much data. And given that a change like ndp had in the last couple of days is very unusual, it's not a bad thing if your model takes time to adjust to that. Changing it now wouldn't make any sense.

    ReplyDelete
  18. Advanced polls being up 35% means very little.

    *Its been trending up election after election
    *It was held on a holiday weekend with nice weather
    *A key part of the CPC's massive GOTV operation is to encourage early voting

    Why people are reading so much into it I don't know.

    Also most of the advanced voting took place BEFORE Layton's massive surge.

    There was hints he was on the upswing nationally and ahead in Quebec going into the weekend but nowhere near he is now.

    Numbers will revert:

    CPC, LPC, and BQ are all higher than they are polling now. CPC espeically so due to GOTV expertise.

    NDP are lower than polls show.

    ReplyDelete
  19. Hi Eric,

    Just wanted to say that I've found your site great reading as the election approaches (and I was a reader even before the writ was dropped).

    I agree with you that too many people are piling on about your model and the like. I think a bit of conservatism with regards to the NDP surge is called for (and I say this as a NDP supporter) and I think it's smart to have a built-in hedge, especially given the problems with polling in recent history. For those who disagree, I have two words: Nick Clegg.

    ReplyDelete
  20. My sympathies Eric.

    You may need a disclaimer on the front page to explain the method if you are to have a hope of putting that talk behind you. :)

    ReplyDelete
  21. I'm not sure that the Liberals are affected in North Vancouver the way they are regionally or nationally. The Conservative incumbant - Andrew Saxton - is strongly disliked by many, and the Liberal candidate (Taleeb Noormohomed) seems competant, intelligent, and well-spoken.

    I will be very interested to see how the student vote plays out here, given the presence of a university in the riding (to which approximately 2,500 students, plus those commuting to SFU/UBC live in), the focus on post-secondary education in the Liberal platform, and the fact that many University of Victoria students will have returned home for the summer in time for the May 2 vote. The Capilano Students' Union held an all-cadidates forum (the CPC candidate had "other committments") where the Liberal was very well recieved, and the NDP and Greens weren't percieved by many of my classmates as being serious contenders. I know for a fact that many student union executives that live in the riding are privately supporting the Liberal candidate in North Vancouver.

    A. Student

    ReplyDelete
  22. People, if you want to see the NDP with 100 seats, just go to the Nanos website and look at the latest poll. If you think that the purpose of this website is to show the latest numbers, you didn't get the point of it.

    Éric, keep up the good work. I was a long time fan of 538.com and was ecstatic when I found that we had a Canadian version of it. I'm simply addicted to your website. I was freaking out because I arrived at work and you haven't yet posted the daily update, hahaha

    ReplyDelete
  23. LOL Eric,
    Your traffic numbers must be skyrocketing with first time visitors. I hope you are actually making some money off the ads to compensate for all the work you are doing. I say they are sktrocketing, because there seem to be a lot of first time visitors complaining that the model isn't telling them what they want to hear.
    Stick to it, while there is some validity to the argument that you fail to capture a snapshot in a volatile environment, (like an election), that is precisely what is intended, no?
    One thing I would suggest though is when you say that the slow movement of the model will end up capturing the difficulties you expect the NDP to have in putting the votes in the bank, you are doing yourself a dis-service, and claiming a credit where it isn't due. Fact is that no poll or model can consistently capture organizational strengths and weaknesses, so an honest poll ( or seat projector) should be up front about the limitations of the model, and recognise that if such strengths and weaknesses are 'captured' by the model, then the model is probably flawed at a more fundamental level.

    ReplyDelete
  24. Sorry Eric, you're doing such a great job that you have a constant flux of new readers who just aren't up to speed on what you're doing. Your model seems pretty sensible, and 2011 looks like it'll give it an excellent test.

    ReplyDelete
  25. @ Tyrone

    If the Orange Wave is real, then there is little need for ground-game. The people will just get out and vote by themselves.

    ReplyDelete
  26. http://www.ekospolitics.com/wp-content/uploads/full_report_april_29_2011.pdf

    New EKOS poll out, new increase in NDP support. I'm willing to take a bet that by Monday night NDP will have reached 32% in national vote and 104 seats....and I might be underestimating their dynamic. I am hearing people who would never ever ever think of voting NDP just a couple of weeks ago, willing to jump on the wagon. Come Tuesday, we will be living in a very different and more optimistic canada...

    Btw, there goes one of the arguments for pro-Harper voting: http://www.cbc.ca/news/business/story/2011/04/29/business-canada-gdp.html

    ReplyDelete
  27. I've been following this site for a long time, way before the election
    took place. I think you are doing a great job Éric and have made me,
    as a young person, much more engaged in Canadian politics, and in
    particular the realities of our electoral system.

    I enjoy reading the comments, but people need to lay off. This is
    Éric's projection and he has stated many a time that he is aware of
    the current polling situation and the limits of his current model.
    It's an experiment and I'm sure tweaking will be required at a later
    time, but thus far its been a successful system and would be
    irresponsible to change without testing it in this new polling
    environment we find ourselves.

    Anyway, thanks for all your hard work Éric!

    Rob

    ReplyDelete
  28. The thing I find interesting is the dramatic rise of the undecided in the poll trend, where I would expect undecideds to decrease as we get close to election day.

    I presume the party numbers are of the decideds. Regardless, it seems that the undecideds are Liberals and Green trying to decide if they're willing to vote NDP. If so, and they decide yes, there could be a HUGE NDP vote - even NDP government might be possible.

    Eric, I'd be interested to see what your model showed if that undecided was allocated to the NDP!

    ReplyDelete
  29. Yeah people need to stop harping on the model. It has a slow response time but come election time it will have reacted quite well. My thoughts are that the real scenario will be between what this model shows and what the polls show.

    ReplyDelete
  30. Nihir,

    Please hold judgment on how well it works until Monday night.

    AverageCanuck,

    You bring up a good point - the increased advance turnout will not benefit the NDP, as much of the surge took place after the weekend. It probably won't be a "game-changing" factor, but it is a factor nonetheless.

    Bluegreenblogger,

    I mean to say that it might unintentionally capture the organizational problems the NDP has in Quebec, note that it will all brilliantly work out as planned.

    ReplyDelete
  31. Eric- Remember that the prevalence of what I would call "inexperienced" comments is generally a demonstration of your success.

    It is the case that there is a lot of excitement (media-fed, of course, but not completely ridiculous) about the way this election seems to be changing, and changing quickly. As a result, if you just read online news sources and twitter feeds, it's easy to come to the conclusion (possibly correct!) that *whoosh!* the results are going to be massively different than what we thought - more than 100 seats for the NDP, or fewer than 15 for the Bloc, or some such extreme result.

    Then, that person comes here for the first time, and sees a race that's not as different as they expected. Not having the benefit of weeks of watching this site, they think that it's some sort of personal opinion of yours that things won't change much.

    So (and particularly if they happen to be a partisan), they post somewhat angrily that you're out to lunch.

    ...which only provides more proof that you've got more new people showing up here every day. If you weren't getting some of these every day, it would largely be evidence that you were no longer attracting new people.

    I look forward to your site on Sunday, and a fascinating evening on Monday!

    ReplyDelete
  32. ็ำั็ำั้ำHey Eric,

    Great stuff, I've enjoyed reading your commentary, and thankful for your efforts. Don't sweat the constant comments on the 'offness' of your model, they are obviously new eyes which didn't have the patience to scroll back a day or two. Your model is what it is, and whatever I may prophesy, or however much polls may differ, I'm keen to see it play out the string.

    This election is great fun for anyone with more than a passing interest in politics. Thanks for making it that much more of a gas.

    Richard

    ReplyDelete
  33. How does your model account for strong independent candidates that won't appear in regional polls? I.e. In the ridings Edmonton - Sherwood Park (The independent is splitting the conservative support with "vote local" campaign); Simcoe-Grey (H. Geurgis is splitting the conservative vote), Saanich Islands (I presume May does a lot better locally than the greens in BC).

    ReplyDelete
  34. Eric,
    does your model still predict that Peter Mackay will lose his seat? Is there any local polling for that region?

    Also NTV poll shows that avalon is a close race in nfld but that the liberals are leading the CPC. Perhaps this should be incorporated into the model.

    ReplyDelete
  35. Just want to say Eric that I appreciate all the work you have done here. I check this site every morning and usually again in the afternoon.
    Thank you for all your hard work.

    MPAVictoria

    ReplyDelete
  36. Eric, don't worry that the model is being criticized. The analysis is what it is. If it is being hit by a once in a generation electoral earthquake - so be it.

    You built a Kepler to look for small changes in a star's brightness and managed to luck into looking right into a supernova. One will still be able to learn from what was is happening. Don't feel compelled to "fix" the answer you are getting. The projection is not Eric's projections, they are your model's. If it proves unresponsive to this sort of a swing we will know soon enough (I'm sure you have already played the "if the next two days show the same trend game" and have some idea of what your model's final projection might be. This was always intended to be a test of a very well considered theory. It is ok for a theory to be wrong, in fact we usually learn more from those situations than the reverse.

    The analysis is very interesting and the conservative, small 'c', projections actually hammer home to me how amazing this election is.

    I would love to know your thoughts on how close the model will be to your personal prediction for seats. I suspect it will be reminiscent of my thoughts during my hockey pool draft come the fourth round: "I don't understand why my lists are lying to me." :-)

    Keep the faith Eric and know that you have done an amazing job here.

    ReplyDelete
  37. @Eric

    Just a quick suggestion:
    Using weighted past polls in creating a current estimate is a good way to smooth out statistical blips, which is good, but it also makes it smooth out actual trends. Presenting both your model and one that weighs the past less heavily (or not at all) would give two useful scenarios where reality is probably somewhere in between. I'd find that really interesting.

    Also, the slow movement of your model is not capturing the problems the NDP face in getting votes out. Your model is slow because of the way it was designed, not because it somehow predicts low NDP turnout. They both bias the polls in same direction but they aren't reflecting the same reality. Sorry, that's the statistician in me coming out.

    All that said, the effort you've put into this is truly inspiring. Keep up the good work. You should use some statistical measures of agreement to see how well your model does against the election results.

    ReplyDelete
  38. Hello? Everyone who's dissing the model? IT'S A MODEL. We don't know how it worked, or not, until there's something called ELECTION DAY, which isn't that far away. Why not just wait and see?

    If you don't like this model, I've got great news: it's a free market! There's another 100 models out there if you want to hand out at a website that already matches your preconceptions!

    I don't agree with all of Eric's thinking by any means, but give credit at least to him for (1) launching an experiment in public where we can see how he's thinking and (2) carrying it through to the end where he can risk all of us shaking our heads at the result.

    In the absence of a result, mocking his model as "wrong" is just as absurd as presuming that one or another of these disparate polls is "right." The answer is, we don't know until we know.

    ReplyDelete
  39. Full disclosure: I'm a die-hard NDP supporter.

    That said, I don't think gaining so many seats in such a short time is good for the long-run prospects for the NDP and for the country:

    What we see in the majority of riding polls is evidence of extreme polarization. Alberta, Saskatchewan and the interior BC are not merely Tory country, they are staunchly so. Out east and on the west coast are congealing into a left-of-centre hodgepodge. In other words, we are becoming 1993 all over again, but this time with different stickers.

    As I said, I'm an NDP supporter -- but let's be honest: we work better as the conscience of the HoC rather than as a governing party. To cite an example, look at the 1990-95 Ontario NDP government, comprised chiefly of activists, protestors, social and labour advocates and educators not used to being the Establishment didn't know how to deal with all this responsibility -- and to deal with a terrible prolongated recession for half the mandate.

    The NDP need a transition phase, and it is my hope that if they do form the offical Opposition, they do so with a period of time so that all the new faces have time to shed some of their green peel (and to hopefully cast off the latest 'outrage' against vacationing candidates (like as if no other party parachutes people into ridings...)

    I'm hoping Éric's numbers are closer to what happens this Monday than some of these wild speculative seat projections from EKOS. Here is my preferred breakdown:

    Cons: 134, Liberals 74, NDP 82, Bloc 16, Greens 2

    The NDP becomes the official opposition, but not too big for their britches. This scenario humiliates both Harper and Iggy (hopefully forcing them to both leave in a few months time) and gives time for the NDP and its newer MPs a chance for more face time with the public.

    ReplyDelete
  40. Do people not understand that just b/c X % say the support a party does not mean that they will get that proportion of seats?

    Eric hopefully the downside of the internet won't hinder your desire to keep this site going as it is a real treat. The armchair statisticians be damned! ; )

    The Orange Tide should be a message to the CPC that it is time to dump Harper if they hope to get a majority. The pendulum swung right and Canadians got scared and now are swinging way left. They need to get a PC at the helm (and the LPC need to support Iggy, to dispell the 'in it for himself' crowd as I don't think they can handle another change at the top) if they hope to ease Canada back to the Conservative brand. It helped fill their coffers having Steve push his weight around but it didn't solidify the top spot. My thoughts anyway

    ReplyDelete
  41. Good show Eric, don't sweat the newbies. As others have pointed out it goes along with being a popular site. I'm impressed with your work and intrigued to see how the model ends up. Thanks for making an interesting election even more so.

    Richard in Bangkok

    ReplyDelete
  42. Eric, I wanted to add my thanks for you doing this. This has been the most exciting election in my 57 year life (except possibly Trudeaumania), and your coverage has been part of that!

    Catch22 and all the other strategic voting and e-engagement sites on the web have helped too, and the coverage in this week's Now articles by Klein and Hollett are worth reading (www.nowtoronto.com).

    When the election was called, I was fairly unhappy. I'm euphoric now... I don't want to contemplate how I'll feel Tuesday if the CPC were to pull off a majority. :-(

    As Klein and Hollett say (differently), these are exciting times. We might be about to transform Canada into a real social democracy!

    ReplyDelete
  43. Eric, stick with it until the end brother. You did way more research than anyone making these comments (and I assume your weeding out 100's of comments as well). If the model doesn't work in the end, it doesn't work, life will go on. I think what you were trying to do was make an attempt to use statistical analysis to try to make a good predictor of the election.

    If your model was like every other model out there (take the latest poll) - then what would be the point? Anyone take the latest poll and pump out a prediction. Why are people on this site if you all you want is the latest results?

    The funny thing is if you end up being right - which I still think you may be as all this NDP hype in the polls does not necessarily mean NDP votes - you will never hear from most of these people again. Most people that get angry at models are because they don't support what they want to see (e.g., NDP dominance), simple as that - keep it up the good work!

    The purpose of a model is not to please people, its to test your own system against the actual results.

    ReplyDelete
  44. "Would it be possible not to have a day of "the model is slow/wrong/ridiculous" comments?"

    LOL, not everyone has been here from the beginning, and not everyone wants to go through the effort of understanding a weighted projection. Keep up the good work, and don't worry about the transient commenters; they'll probably not be here after Monday!

    Just be happy that you're as popular as you are; you're obviously providing a service that people want!

    ReplyDelete
  45. Eric: You're right, the "new" party has far weaker machines, and its newer fans are particularly susceptible to changing their minds at the last minute. The disappointment that greeted the Liberal Democrats last UK election night is still a possibility for the NDP. Me, if Gilles Duceppe loses his seat I'll cheer, but I'm not going to completely believe it will happen until it happens, no matter how much I want to.

    That said, I'm afraid the critics have a point. A slow discount rate for old polls may make sense in situations like 2006 or 2008 when vote shares are fairly stable. It may make less sense when potential changes in vote shares are large (as in 1984 or, few still dispute, 2011).

    If it's any comfort, at least one other vote share model I know of (at the Polling Observation Blog) still has the Bloc at 9.5 percent! It also uses data from past elections where the Bloc vote was more stable, and can't easily accommodate large changes in the vote share as anything but measurement error (i.e. the polls are wrong). It has captured the rise in the NDP vote fairly well, but practically all of that gets stolen from the Tories and Grits (POB has the Tories down to 33.3 percent), not the Bloc.

    Will you have any time before the election to check (and report) the sensitivity of the results to the speed at which you discount old polls?

    Richard Slater

    ReplyDelete
  46. I think your model is doing the right thing Eric. A surge late in the game isn't always going to translate into votes on Eday. Not to mention that people have this fun tendency to lie to pollsters. Be it to hide their true voting intentions, or just to mess with them for calling at dinner time, it happens.

    There is the bandwagon effect, but come Eday, I have a feeling that some people will calm down and vote normally; or just not vote. (which is ~35-40% of the potential voters; mostly NDP potential votes) Then there is the whiplash effect. If too many undecideds see too large of an NDP 'wave', they'll jump to a party that will stop it.

    ReplyDelete
  47. Harper is toast. The pic in the Star that goes with it shows a face that is of a defeated man.

    Robert Benzie
    Queen's Park Bureau Chief
    NIAGARA FALLS—Stephen Harper's Conservatives must win 23 more seats in Ontario to achieve their coveted majority, a task that senior party insiders now admit is almost impossible, the Star has learned.

    High-ranking sources confide that even with the collapse of Michael Ignatieff's Liberals — and NDP Leader Jack Layton's surge, which helps split the vote in many Ontario ridings — it will be very difficult to make such immense gains in Canada's most populous province.

    At the dissolution of Parliament, the minority Tories held 51 of Ontario's 106 federal seats.

    Party sources say the possible loss of several British Columbia ridings to the New Democrats — and others in Quebec, where Layton is surfing an orange wave — has forced them to revise their projections.

    ReplyDelete
  48. There is always a difference between what people say and what they actually do. Polls are inevitably subject to this human vagary and are always off a bit. In this case we have a tectonic shift of opinion that models will eventually catch over time, but at the present time models will lag the shift of opinion because of the noise filters built into models that discount single outlying polls.

    ReplyDelete
  49. @ nihir, you said "There's clearly interest in a projection, but I think everyone would appreciate accuracy in the model."

    WHAT ACCURACY?! The results have not happened yet, so how can any person out there be looking for "accuracy" in projection models!

    I suspect all of these people are over zealous NDP supporters - NDP supporters are becoming what they despised about the Con supporters - which is turning a blind eye to everything their party does and becoming arrogant.

    "Brosseau is reportedly an employee of a pub on the Carleton University campus in Ottawa. The Globe and Mail reported that Brosseau’s boss didn’t even know she was a candidate.

    Now comes the realization that Brosseau cannot speak French — or barely can."

    There are NDP supporters in the posts responding to that story someone defending this?!

    This proves no matter what party, Lib, NDP, Con, Bloc, etc. - they all contain some crazies. I know NDP likes to their their stuff doesn't smell but it does.

    ReplyDelete
  50. It's no longer time for Éric to change his model. I see he's actually done a few tweaks some days ago, which is the most he can do for now. This election could lead to a complete realignment of the electorate, like in 1993 (which I have no idea how we could model) or it might not. Only on May 3 will we know, and at that point Éric will look at the model, see what worked and what didn't, and make the necessary changes. Until then he has to stand by his predictions.

    The "last polls digest" he'll post on Sunday may give us a better idea of what the result will be, but we should also remember that being based on few polls, it could also be more volatile.

    ReplyDelete
  51. Nelson T:

    About advance polling: it's hard to quantify how much of the Tories' GOTV program have effected the early voting, because there has also been a very successful national campaign to get post-secondary students out to vote on those three days. Many of those students are voting for the first time, and they are more likely to vote for anyone other than the Conservatives.

    I do think voter participation being up is a sign of something: people are fired up about this election, either to preserve the status quo or to make changes. Advance polling in Alberta probably means something completely different than in Vancouver or in Québec, though.

    High turnout typically benefits challengers and incumbents of smaller parties. I see no indication of the contrary.

    ReplyDelete
  52. if the NDP numbers don't translate to more seats, it is clearly because of our very poor-un-democratic electoral system.

    we really need proportional representation.

    ReplyDelete
  53. You
    Gotta Love This !!






    The NDP's Jack Layton is now the national leadership
    frontrunner after surpassing the Conservative's Stephen Harper in a
    pair of key indicators, new poll numbers suggest.




    With less than a week before the election, Layton's personal
    surge comes as his party rides a wave of support that has seen them
    leapfrog over the struggling Liberals for second place nationally.




    While the Conservative party still holds a slight lead in
    terms of overall party support, the NDP has thinned that gap to only
    six percentage points this week, according to poll numbers conducted
    for conducted for CTV and the Globe and Mail.

    ReplyDelete
  54. The daily averages graph shows a virtually straight line slope of increasing NDP support since April 19 to today. IF that slope continues to May 2, there MIGHT be an extra 4% for the NDP. Some technical analysts researching stocks would happily project that number to 100% support 70 days from now, but as Newton observed, what goes up comes down. This number could flatten or go down in the next few days. In the FPTP system, very small %-tage shifts can translate into several seats. Tuesday will tell.

    Eric, thank you for your fine work.

    ReplyDelete
  55. In your methodology you refer to "New Star Candidate" and "Incumbency" as factors in your projection. It seems like those factors did not apply for Charlottetown riding. I am not sure if you care and have time to deal with each specific riding, but if you do here are facts about this one:
    - Last couple elections Liberals were going with super popular candidate. He had my vote. He is not running this time around
    - Three elections ago NDP had 30% of vote here. NDP did not run any meaningful campaign while the above Liberal was incumbent, so they had rather low numbers on last two elections.
    - New liberal candidate comes with a baggage and stays out of sight.
    - Local NDP is running full campaign and have by far the most media coverage here.
    - New NDP candidate is well known and very popular in this community.

    I do not want to get into discussion on how polls apply to your model. Just based on the facts on the ground, with no consideration for poll results at all, your numbers do not stand.

    Sasha

    ReplyDelete
  56. Plot the error bands, PLEASE !

    the little upticks and squiggles are MEANINGLESS without the error bars plotted.

    It is not hard to make these plots with blue, orange, red, etc error bands on them showing the likely margin for error. It not only makes the plots more accurate, but much more meaningful.

    Basic, basic first year data analysis stuff.

    ReplyDelete
  57. Thing is, if 3-5% of the NDP vote doesn't materialize, and I think that's a fair estimate, and if the Conservative vote is under by 2%, everything changes.

    The riding breakdowns for the advanced polls show it was the Conservatives who got the vote to the advanced polls. i live in an NDP riding. I voted advanced. There was nobody there.

    They will definitely have more people vote for them. As many as the polls say? I doubt it.

    ReplyDelete
  58. Fascinating thing with today's numbers: yesterday you pointed out 12 seats where the NDP were within 5 points or less in the projection. Today, they're threatening in 22 ridings (17 in Quebec alone). By my count (and admittedly I just got up), if all those seats were to change hands, we'd see: CPC - 138, LPC - 67, NDP - 75, BQ - 28.

    ReplyDelete
  59. Eric, thanks again for all your work and detailed explanation of your model. It's the best model by far that I've seen. That said, I don't think any model can accurately capture what's been happening in the past week or so. Chances are somebody will predict it right -- if only by luck -- but if it's not you that doesn't invalidate the model. I doubt we'll see another election like this one for a while.

    If I could suggest one tweak to test when you're doing any post-election analysis, perhaps if overall trending is indicating rapid changes (as we've seen) your weighting factor could adjust to reduce the impact of older polls more quickly (e.g. 14% per day instead of 7%).

    Just a suggestion - not a criticism. Thanks for the enjoyable reads.

    ReplyDelete
  60. David,

    Strong independents like Arthur and Ford were given a certain percentage of last election's vote, based on how independents like them have seen their vote drop in the past. For Guergis, her vote was taken from the Conservative total based on how other MPs-turned-independents have seen their old vote shift over in the past.

    ReplyDelete
  61. Do you think you could produce both a nationwide and regional trend graph of your seat projections by day, similar to your ones for popular vote? It'd be interested to be able to compare the two.

    ReplyDelete
  62. For all the hoopla surrounding thee NDP surge, it's barely nudged the Liberals and Conservatives outside the margin of error of earlier polls. Just look at today's (April 29th) seat projection for the Conservatives.

    I'm not sure if the NDP platform would stand up to a lot of scrutiny of the campaign were to last a few more days. Also worrying is their pledge to apply Bill 101 rules to Federally regulated workers in Quebec. Not good. Bad for equality and democracy.

    It'll come down to the ground work on Monday.

    It'll come down to

    ReplyDelete
  63. And we have a winner! The EKOS poll today shows among those absolutely certain to vote the Liberal Party is at 19.9%.

    That is something they were hoping against hope not to see.

    ReplyDelete
  64. The model is fine. It has weaknesses and strengths just like other models. The important thing is to be consistent.

    Unfortunately, Eric had to change the model once mid stream this election.

    Projections, as well as polls are most useful for seeing trends over time as compared to themselves, not other pollsters or other projection models.

    We can clearly see the trends of the past few weeks in Eric's model.

    No model can take into account all riding level factors, such has individual candidates, levels of organization, local issues, ethnic votes etc. These will be crucial on election day.

    What an overall model assumes is that an equal number of these factors over the entire country will favour each party proportionally.

    I am not convinced that is true this time. The Conservatives are much more organized than the other parties. The NDP the least so. In certain ridings, the Liberals are also very organized. The BLOC is fierce.

    Sure the NDP vote has increased, but will it all materialize. Very unlikely. Well organized Liberal ridings, with well liked incumbants may still deliver for the Liberals. Vote splits may give a few more to the Cons in Ontario. The Bloc will DEFINATELY hold on to more than the 3 or 4 seats some projections are calling.

    NO model can deal with the organization factor. It can only show polling trends. That's all. Eric's model does that. He should absolutely not change it now or it would lose its value.

    ReplyDelete
  65. Well looks like the National Post just did a hackjob on this site...
    http://fullcomment.nationalpost.com/2011/04/29/alice-funke-dont-take-seat-projections-seriously/

    ReplyDelete
  66. Quebec City poll today :

    NPD : 34%
    Conservatives : 27%
    Bloc Q : 27%
    Liberals : 9%
    Greens : 3%

    ReplyDelete
  67. Hey, great model. Examining on a riding by riding basis the projections seem fairly accurate. A significant number of seats in Quebec are about to change to the NDP so a big change could be ahead tomorrow. Anyway keep up the good work man :)

    ReplyDelete
  68. "With respect, if you continue to post updates using a model you admit is flawed, you can't be surprised when people call you on it."

    I don't think Éric does admit that his madel is flawed. I certainly don't think his model is flawed. His model is conservative, which makes a lot of sense given past election results.

    ReplyDelete
  69. Eric's current predictions, while likely underestimating NDP seats in Quebec are spot on for the Conservatives according to Frank Graves latest projections.

    The Liberal and NDP vote will split Ontario and Harper will come right up the middle and probably capture a majority of the Ontario seats. The possibility of Layton handing Harper a majority is still very much in play if Ontario continues this split.

    As a Liberal supporter myself, I consider this the "nightmare" Harper majority scenario.

    http://www.ekospolitics.com/index.php/2011/04/from-orange-crush-to-red-crushed-new-ontario-vote-splitting-favours-conservatives-april-29-2011/

    ReplyDelete
  70. Hey I'm curious about one thing. Your model is showing Laurier - Sainte Marie as being a virtual toss-up between Bloc and NDP. This is Gilles Duceppe's riding. I would be surprised if it's that close. What are those numbers based on for that riding? Does your model account for the fact that a party leader is sunning in a riding?

    ReplyDelete
  71. If this orange surge is real and holds, than progressives should be grateful to Layton and the NDP for stopping Harper, something the Liberals couldn't do; and all Canadians should be grateful for the demise of the BLOC and the prospect of re-engaging not only Quebec, but all parts of the country, something the Conservatives and Liberals couldn't do. Maybe all Canadians, in their collective wisdom, want this great country to heal, choosing Layton and his party .. warts and all! A leader himself healing from prostate cancer and hip surgery, showing great energy and optimism. The political hacks can do their spin, but it is clear more than 60% of Canadians don't want the Conservatives. Sorry for the rant. Canadians will decide May 2nd. Great work Eric!

    South Parkdale Jack.

    ReplyDelete
  72. "Anonymous said...
    if the NDP numbers don't translate to more seats, it is clearly because of our very poor-un-democratic electoral system.

    we really need proportional representation."

    No we don't. PR unfairly helps Southern Ontario and Quebec while leaving the Atlantic and most of the Western provinces out in the cold. If we really wanted to reform our system we should either do MMP with a fairly small number of list MP's that would simply regulate some unfairness while keeping constituencies or we should change to an instant runoff systen which would work best because an MP could not be elected unless they had 50% of the vote and people could vote for smaller parties if they wanted to without fear of wasting their vote.

    ReplyDelete
  73. Eric;

    I greatly enjoy your site and, while having my own strong political views, would encourage you to - within reasonableness - stick to your model. The need has never been stronger for evidence-based decision-making, and your approach embodies that.

    People who accuse you of not being accurate or simply don't like your predictions are missing the point. Worst case scenario is that you'll be wrong for the right reasons, and the next kick at the can will be even more refined.

    My feeling is that the very black swan nature of this election will ultimately do wonders for the model :o)

    ReplyDelete
  74. Gar Y

    Eric this is good stuff. I have watched in this campaign talk about voters wanting stability and Harper just needing to focus on the economy. He did what he was supposed to in the debates etc and since he was blessed with the most solid supporters in the polls it looked good. His strength: worry about the economy and a solid base unwilling to switch their vote.

    What happened? Layton got off a few great one liners in the debates and suddenly all the above doesn't apply anymore.

    No I don't think so.

    So should we now conclude the drivers above suddenly mean nothing and those bedrock voters suddenly fall in love with a person they have seen for years. NO.

    What we have here is a Quebec Tide ready to flow in some other direction and found it in a personality that they may not have really known much about. As Harper,Mulroney Iggy, Bouchard, Gilles learnt you don't exist in Quebec until Quebecers say you exist. And then they speak in ONE voice. I think it is almost aristocratic in nature, blessed by few then required of the many. How else can you explain artists switching the vote in 2008.

    I don't believe what had underlined support for Harper (strength in party loyalty and stress on good management) disappeared in the ROC. It was never a strong sentiment in Quebec anyway from what I saw. I do believe it is still a prime driver in the ROC, just overwhelmed by an exciting media story and will be a rock the Orange will not be able to Crush by Tuesday.

    Again I point out there is only significant evidence in the numbers that what everbody is dreaming about is occuring only in Quebec and that this is driving the NDP national numbers.

    In the other regions the numbers are really fluctuating within their margins of error. The right number to watch again is only the national one. I believe in the ROC the CPC loyalty and good managers drivers are still in play and buffetted the Orange Wave.

    The National Numbers show remarkable stability for the CPC (but I must admit at minority only level now) with LIB switching with NDP because of what is happening Quebec. I believe Nationally the polling numbers for the CPC will return to where they were on April 1st but at a new national configuration that will lose them seats in Quebec and BC.

    There actually seems to be some evidence of this return to Status Quo in the most recent EKOs poll (the poll that first broke the Orange Crush open). Nationally the CPC see some improvement.

    So therefore, I believe the approach here is exactly right - reflect all factors not just the latest fad. Folks - this is Eric, not Glenn Beck.

    My predictions are:

    1. Whatever Eric finishes with will be very close to the results
    2. Quebec will see a lot more NDP, fewer CPCs and LIBs and finally a coffin for the Bloc
    3. Ontario will see more CPC
    4. BC fewer CPC

    I like what someone else wrote 138 CPC and 16 Bloc. With this math does it really matter how the rest go. Everybody get what they want but Harper and even he should be happy with the demise of the Bloc and major step toward his primary goal of wiping the Liberals off the face of Canada.

    And even I a Red Tory for my party!

    ReplyDelete
  75. Eric:
    thanks for your excellent work.
    based on events of the last week,
    you also need to include a second forecast based on a model of current polls only:
    Harris-Decima April 28: CPC (35) NDP (30) LPC (22) BQ (5)
    EKOS April 28: CPC (34.8) NDP (27.5) LPC (22.3) BQ (6.1)
    Nanos April 28: CPC (36.6) NDP (30.4) LPC (21.9) BQ (6.0)
    308.com April 28: CPC (36.9) NDP (22.9) LPC (25.4) BQ (7.5)

    ...Seat Projection:
    EKOS April 28: CPC (139) NDP (98) LPC (56) BQ (14) Ind (1)
    LISPOP April 27: CPC (147) NDP (69) LPC (60) BQ (32)
    308.com April 28: CPC (143) NDP (47) LPC (74) BQ (43) Ind (1)

    and even these polls and seat projections are stale.

    do your regular weighted projection and also a second one that peels off the weight of polls of previous days...maybe just do a band of numbers.

    All the best.

    ReplyDelete
  76. "Honestly: I was very sympathetic to Harper when the campaign began, but now I would vote NDP only for the thrill of it, for change, for fun, to see what will happen, to be also a part of the Orange Wave."

    This attitude is reckless. Will we have the bar waitress who doesn't speak French for Minister of Human Resources? Please think about what you are voting for. Their platform is uncosted, unrealistic and made of fairy dust.

    Other than about 50 candidates, their people are post holders i.e. non-candidates. If you think corruption, incompetence and scandal were bad under the Conservatives, you ain't seen nothing if these people actually get elected.

    We will also have another election in less than a year if the NDP even get to opposition status and vote down the Cons hoping to gain power. Then their numbers will be decimated in Quebec, when people realize their MP's don't even speak French and don't live in the province, or are the local meat cutter in Ontario.

    An NDP government is a train wreck waiting to happen. I'm sure that Jack doesn't even want that, because he knows the ultimate outcome a year form now.

    I urge people to take a second look before they vote on Monday.

    ReplyDelete
  77. So the NDP has about 50 potential MP's worth thinking about? Sounds about right. Since many PM's have said the cabinet is all that matters and the rest of the MP's are trained seals (forget which said it, could've been pretty much any PM who lasted more than a year in the past 40 years though) who cares if they are 'a bar waitress who doesn't speak French'?

    Now, if MP's suddenly had power things would be different but I seriously doubt any PM would do that.

    ReplyDelete
  78. I really appreciate this site -- I check it more than once a day. Thanks for all the work you put into it!

    ReplyDelete
  79. Absolutely, Éric. People who have followed other projection sites (like 538 and Electoral-Vote.com) understand that this is a model, not just the latest polls. People open up the EKOS poll and dance, then come here and don't understand it, don't take the time to read the methodology.

    I'm new to the site too, but I am thoroughly enjoying it. Keep it up, and I look forward to seeing how the model gets tweaked next time out as well.

    ReplyDelete
  80. "Anonymous said...
    "Honestly: I was very sympathetic to Harper when the campaign began, but now I would vote NDP only for the thrill of it, for change, for fun, to see what will happen, to be also a part of the Orange Wave."

    This attitude is reckless. Will we have the bar waitress who doesn't speak French for Minister of Human Resources?"

    You do realize that the House of Commons is the house of the people to be represented by the people? That fact that someone is running and could potentially win that is not a lawyer or a doctor is actually really refreshing. Also someone doesn't have to have a phd in the field that they become minister in. That's why there are entire departments of people that work in a special field, to advice a minster and to do the real work. A minister simple sets the course.

    You are really just showing your own lack of understanding at what the point of our democracy is with statements like that

    ReplyDelete
  81. Nelson T. --

    http://threehundredeight.blogspot.com/2011/04/new-democrats-up-liberals-and-bloc-down.html?showComment=1304099650323#c9024024673824582945 "An NDP government is a train wreck waiting to happen. I'm sure that Jack doesn't even want that, because he knows the ultimate outcome a year form now.

    I urge people to take a second look before they vote on Monday."

    People have been taking a 'second, third, fourth, even fifth look for about 5 years now. Many of them find it hard to imagine any party right now doing worse on ethics, governance, cronyism, democracy, and most of all, the economy than these Conservatives. Goodness help us if they ever get a Majority.

    One thing remains clear, regardless of anyone's polls or projection: the Conservatives have their base, and right now, its base is large enough to fill a Minority government. And it will be that way for a very, very long time. At least until the oil sands dry up.

    I take exception to the person who said the NDP are becoming like the people they despise. Most NDP supporters don't despise anyone. They just want things to change.

    ReplyDelete
  82. Too bad about the comments.

    I suspect the hope for change that lies behind the NDP surge lies behind the desire to see this particular projection site confirm the surge. Its really really difficult days for an NDP supporter, when one thought is of an NDP led coalition and another is of a squeeker Tory majority through vote splitting all the while worrying about whether the vote will get out there while being optimistic.

    Still, nice to see hope triumph fear in a Canadian election. We havn't done that as a nation in a long time.

    Maybe consider next time, and please keep doing this work, putting a link directly under the numbers saying "If you think the overall seat projection is wrong, please read this." There you could do a one paragraph bit about the nature of the projection and how it reacts to change. Might save you a bit of aggravation.

    ReplyDelete
  83. The last two weeks of this election has been about the NDP surge. The media has been every enthusiastic about the phenomenon and I'm not surprised giving the stale tone first half of the campaign. The common consensus was that nothing will change in the house of commons with this election.

    Now things are much different. Even the most modest seat projections like this website show crucial NDP gains that will change Canadian politics. I doubt the NDP support would crash by May 2nd, but it might go down a couple of points.

    My personal prediction of the final outcome is
    Conservative - 37%
    New Democrat - 27%
    Liberal - 24%
    Bloc - 6.5%
    Green - 4.5%

    It all depends on how the vote will be split throughout different parts of the country. Either way Monday night would be more interesting than people would have thought a month ago.

    ReplyDelete
  84. People criticizing and defending this model are doing so based on the exact same reasoning. It's slow so it's not sensitive to blips in the data (which people love) but it also means it down-weighs and is slow to reflect actual trends (which people hate).

    If there is an actual linear trend in a party's polling up to election day, I would expect the last day's polls to be a better estimate of the outcome than a weighted average of polls over the campaign. If, on the other hand, a party's numbers are relatively stable, then the weighted average is probably best.

    One thing that would help with this site is to differentiate seats with a clear front runner from those who are simply leaning one-way or in a toss-up. Also some kind of boot-strapped error bars would be nice.

    ReplyDelete
  85. "This attitude is reckless. Will we have the bar waitress who doesn't speak French for Minister of Human Resources?"

    That only worries people who believe that all politicians should be attorneys. Perhaps she won't work out as an MP, and she'll get shuttled off during the next election. Perhaps she'll turn out to be fantastic, and she'll have a career in politics that spans decades. In either event, it's impossible to know.

    ReplyDelete
  86. I think you have a point re: GOTV issues for the NDP. But I think you are overlooking a number of ridings where there has been previous NDP strangth, i.e. - Beaches-EY, Parkdale and Davenport. I know it Toronto centric, but it's the only ones I know well. In each of those there is a strong possibility of established NDP organization to capatalize on the "orange crush".

    ReplyDelete
  87. People who are blaise about a bar waitress who doesn't care about politics and hasn't campaigned and was out of the country until recently, and who doesn't live in the province far less riding she is meant to represent, becoming an MP, have absolutely no idea how Parliament works.

    If she were a bar waitress with a degree or diploma who at least cared about politics, who lived in her riding and had campaigned for her riding, maybe she could sit on the bench. She would have to care enough to learn, to show up, to have an office in her constituency, to go to endless committee meetings and be there for her constituents for it not be be a completely reckless vote.

    She does not care enough. Does not know enough. Does not even live in the province of Quebec, far less riding. How is she to represent her constituents without speaking their language? It is NOT ok, not a joke or for fun. Voting to feel a part of something with no regard for what may happen is an abuse of democracy.

    If people think Ignatieff's 70% absenteeism is a problem, even when he was working for his party elsewhere, and it is, then not being available at all during a campaign, and we can assume future government, should be as well.

    This is not a waitress who cares. This is a waitress who doesn't care. Big difference.

    ReplyDelete
  88. Gotta agree with Og Eric. Super job and just ignore the snide stuff.

    Thanks for your efforts, appreciated

    ReplyDelete
  89. @Anon@ 15:14

    "People who are blaise about a bar waitress who doesn't care about politics and hasn't campaigned and was out of the country until recently, and who doesn't live in the province far less riding she is meant to represent, becoming an MP, have absolutely no idea how Parliament works."

    I actually have a pretty darned good idea of how Parliament works, and how the bureaucracy works. There's a reason why any position of any importance includes a number of permanent bureaucrats who are there to inform and guide decisions. Even back-bench MP's are supported by staff who are chosen to ensure that the MP is informed, organized, and effective.

    So she was sent to Parliament based on no action of her own. This indicates that her constituents so disliked their other choices that they would instead vote for a complete unknown. And what it means is that she will have to work herself to the bone if she ever wants to be re-elected in her riding.

    She got one free pass, and the rest she'll have to work for. And, quite frankly, the people of Berthier--Maskinongé are probably going to keep on living just the way they always have. Perhaps, they might even be pleasantly surprised.

    ReplyDelete
  90. Enough already about the waitress! What they use to say years and years ago about the Conservatives on the East coast ... that they could run a dog and it would win, and they often did!

    South Parkdale Jack

    ReplyDelete
  91. Amateur Psephologist29 April, 2011 16:15

    Eric

    As I have said, I am greatly delighted with this site.

    At the same time, I am sufficiently respectful of your motives and your standards to believe that you can benefit substantially from people who will disagree with you and your model, sometimes sharply. Why?

    1) A site where debate and contention are rife is a live site, not a dead one; people will keep coming for that
    2) While people may say things you find critical of you, as long as they are not being derogatory, you can learn from their input - don't we all believe in the 'Wisdom of Crowds'?
    3) If your supporters up above are right, and most of the critics will be gone on Tuesday, never to return, that would be a big loss and not just in eyeballs for advertisers. Now is your 'moment of truth' - an opportunity to demonstrate to the crowd that this is a great place to debate and, yes, learn neat stuff

    So, ironically, I found your plea to lay off the criticisms of your model eerily like that of Jack Layton yesterday. He was being deluged with quesitons by all sorts of rude reporters who wouldn't let him finish his sentence before yelling out the next one. He seemed really irritated, and I can see why.

    But still, he had a chance to show how pleased he was about this evidence of his success. (Just remember that, like Layton, you are winning...)

    ReplyDelete
  92. Thanks for keeping up all the old seat projections, Eric, it's appreciated.

    For those who don't think this model reacts very well, compare two weeks ago to today.

    http://4.bp.blogspot.com/-cI24jpn-ULg/Tbq-FygL1eI/AAAAAAAAFAM/UEWQwFGIV3o/s1600/11-04-29+Ridings.PNG

    http://4.bp.blogspot.com/-mCpvPuRPQgs/Tag-6_h7dfI/AAAAAAAAEwM/v4e5lAiq9Xs/s1600/11-04-15+Ridings.PNG

    Then compare to the current NDP/BQ split in Quebec and you can get a feeling for how far many ridings have come so far. When you think that there's still a 15 point spread that may not be fully incorporated into the model, it shows just how powerful the NDP surge could be on Monday night. Really quite fascinating.

    ReplyDelete
  93. Oh, forgot to mention earlier to everyone that Orange Crush is on sale at No Frills in Toronto. What does that mean????!!!! Sorry Eric.

    South Parkdale Jack.

    ReplyDelete
  94. Think everybody is making waaaay too much of the 34.5% rise in advance polling.

    The rise in ALBERTA was 34% as well.. just about dead on the national average.

    Out of the 28 Alberta ridings, maybe, and I mean maybe, there are 3 seats that are not absolute dead Tory safe.

    Edmonton Strathcona the NDP likely holds, in Edmonton Centre the Grit might have had a shot if Iggy hadn't imploded, and there is a strong "independent Conservative" spitting the vote in Edmonton east so (maybe) that seat could be close....

    The other 25 are about as exciting as watching paint dry...

    It was a holiday weekend, we had some nice weather....

    To quote Freud... sometimes a cigar, is just a cigar....

    The Vorlon

    ReplyDelete
  95. I'd be interested to see historical data on how predictive of voter turnout the advance polls are. Without some sort of context, that 34% increase tells us nothing.

    ReplyDelete
  96. Would be interesting so see the ceilings this week. Some really high numbers that come to mind for the NDP:

    42% as per Harris Decima in Quebec 20 points ahead of the Bloc and 25-30 ahead of Conservatives and Liberals
    40% in Man/Sask as per Forum, tied with CPC
    35% and clearly in 1st in the Atlantic as per Forum and Ekos
    29% in Ontario as per Nanos in Ontario, tied with the Liberals and only 7-8% behind the CPC
    25% in Ontario with the Conservatives and Liberals at 34 and 33 as per Harris Decima
    31% in BC as per one EKOS poll only about 4.5% back of the CPC
    20% in Alberta in Ekos, 37 points behind the CPC, but much closer than 52 points in 2008, could tip Edmonton East to them and Edmonton Sherwood to the Independent candidate.

    There must be an ideal scenario where that gives them a minority government at least.

    ReplyDelete
  97. @Amateur Psephologist

    I am not speaking for Eric here, only me but at least me, comments like suggesting that the model put in a factor for aggressive last second trends is helpful and adds something to the site and the projection.

    But the comments like how this projection is not "accurate" make no sense as the results havn't happened or comments like "I just saw the EKOS projection, your way off!" are annoying. They don't help anything except criticize a process people don't understand or havn't even tried to understand.

    But agreed, any comment that provides suggestions or insight into improving the model are great - at least from my perspective.

    ReplyDelete
  98. Brian from Canada29 April, 2011 17:14

    The amount of invective here against so called post holders is truly saddening. The Conservative supporter, and it is obvious that you are, who is leading the charge on the comments is either a political operative spinning a yarn or naively ignorant about electoral politics.

    In this campaign, we will see over a 1000 candidates from the five major parties and most of them could be called post holders. That would include about 50 plus Conservatives in Quebec. Just because they do not live in the riding does not necessarily mean they have no skills. The comments denigrating the "waitress" are truly sophmoric. She is a polysci student at a university, active in student politics and has worked in many elections. She is also a waitress working to put herself through school.

    Now, to some, being a student or a waitress, may disqualify a candidate, but if that were true, our current PM started his career as a paid lobbyist for a radical organization. Which is more questionable? The answer is neither.

    Candidates come in all sizes, shapes and have varied skills and histories. The one thing they have in common is that they have chosen to participate and put their name forward. Rather than condemn that, we should applaud their commitment to the country.

    Cheap shots have no place on this site and if you can't control your passions, I suggest you go elsewhere and let those of us who see the real value of this site enjoy it without your childish antics.

    to Eric, keep up the good work, you have made this election far more interesting and enjoyable for many.

    ReplyDelete
  99. New Ipsos Reid poll:

    New IPSOS - CPC 38% ... NDP 33% ...LPC 18%...Bloc 7%... Green 4%. In BC, 42% CPC, NDP 29%, LPC 26%

    ReplyDelete
  100. Come on people, I'm with Éric here. How many times do we have to point to the Nick Phenomenon to realize that it makes sense to be cautious here. Most projections that rushed ahead on the UK election ended up giving the Lib Dems nearly twice as many seats as they actually received, and it will likely be the same for those that are throwing a hundred seats at the NDP now. Yes, game changers exist, but not every upswing in the polls turns into a victory at the polls.

    ReplyDelete
  101. @Brian

    Sorry but I don't think its unreasonable to expect the following of our candidates:

    1) They live in the riding or have a close connection to it

    2) They speak the language of its residents so as to be able to represent them in Ottawa

    Bonus 3) They spent at least some time campaigning or at least stayed in the riding during the election ...

    ReplyDelete
  102. She's not a student at all in Poli-sci or otherwise, or active in politics at all. Her bio says she has helped find homes for lost animals in her neighborhood. No one who knows her has heard her mention politics.

    You are confusing her with the NDP student club presidents who are running in ridings they have never visited, which is also unacceptable to vote for. Anyone who votes for someone to represent them in parliament who has never even visited the riding, far less campaigned in it, or lives in it, is the one being ruled by pure emotion and a lack of reasoning.

    ReplyDelete
  103. "Anonymous said...
    She's not a student at all in Poli-sci or otherwise, or active in politics at all. Her bio says she has helped find homes for lost animals in her neighborhood. No one who knows her has heard her mention politics.

    You are confusing her with the NDP student club presidents who are running in ridings they have never visited, which is also unacceptable to vote for. Anyone who votes for someone to represent them in parliament who has never even visited the riding, far less campaigned in it, or lives in it, is the one being ruled by pure emotion and a lack of reasoning."

    No I'm not confusing her at all. I know who you're talking about and I think that if a normal person wants to run to represent people then all the power to her. And just so you don't confuse my intentions and try to make me say something I'm not saying. I'm not an NDP supporter. I'm actually a member of the Conservative Party. I believe though that the point of the lower house of parliament was to be the people's house while the upper house was meant for the political elite. We seemed to have forgotten that and for some reason thought that people who simply have convictions are not able to make right choices. I think that's the wrong way of looking. Political leaders have not been able to stir people up in a very long time. Maybe it's time that people started taking an active role and saying "I want to run for politics, not with a law degree or a phd but as an average middle class Canadian" I think that's respectable and there should be more people like that.

    As far as people who do not live in the riding and have never visited the riding I think there is a problem there. That (which ALL parties do) spits in the face of grassroots democracy

    ReplyDelete
  104. Anonymous said...
    "If we really wanted to reform our system we should either do MMP with a fairly small number of list MP's that would simply regulate some unfairness while keeping constituencies or we should change to an instant runoff system"

    Mixed Member Proportional Representation was the form of PR recommended to Parliament by the Law Commission of Canada. Additionally, you could use IRV for the local constituency seats, allowing you to have the best of both worlds as well as serving to reduce the number of list seats you would need in such a system, and IRV would make your local seat counts closer to the PR adjustment goal already.

    ReplyDelete
  105. Your site's final projection is interesting and close to what many others predict. I've been surfing and seeing all sorts of stuff. But I pay attention to what is within the margin of error. I find Ekos interesting as even NDP friends are saying their numbers are nearly exactly the same as threehundredandeight.
    I find Ekos more exciting as it predicts a strong NDP outcome. Others tell me they aren't considering the power of incumbents.
    Here's what Ekos's final numbers are
    "After the ballots are counted tomorrow, we expect to see the following:
    1) CPC: 130 to 146 seats
    2) NDP: 103 to 123 seats
    3) LPC: 36 to 46 seats
    4) BQ: 10 to 20 seats
    5) GP: 1 seat"

    ReplyDelete
  106. It has been an exciting election because it has seemed anything is possible. In fact, it has never been more clear that an election campaign is designed to get a party (re)elected and absolutely NOT to debate policy issues and ideas in the light of day. We need to maintain multi-party choice in Canada and we need more professional journalists shining light in the dark corners.

    ReplyDelete

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.