Wednesday, October 21, 2015

First sifting through the results

A big election with some big changes, so there is a lot to go through. I did my first overview of the results for the CBC here, so I suggest you start there.

Here's the latest episode of the Pollcast, and the last to deal directly with the results of the election. It's a good one. Joining me is the CBC's Catherine Cullen, Tom Parry, and James Cudmore. They were with the three campaigns in the final days, and we talk about what they were seeing.

Coming in the coming days (potentially into next week): a review of how the polls did, a review of how the projection did, and where we go from here.

Some initial thoughts, though. The polls ranged from a pretty good job to a terrific job, so I think they deserve full credit. Seat projections across the board, though, were off. And I think there is something to learn from that, as the vote results should not have delivered a majority government to the Liberals. Their vote came out disproportionately in seats they could win. I think that tells us something.

Covering this campaign with the CBC was one of the best experiences I've had. It was a fun campaign to cover. For regular readers of this site, I know it was different from how it has been in the past. But thanks for nevertheless coming to this site every day and leaving your comments.

Throughout the campaign, this site received 7.1 million hits. Suffice to say, that smashed the site's previous records set during the 2011 federal election campaign. I hope many of you will return to the site and stay with it as we turn our attention from the election campaign to the fate of this majority Liberal government. And until 2019, we should have some interesting leadership races and provincial campaigns to cover too.

63 comments:

  1. Thanks, Éric. I, for one, will keep coming here. Good job.

    The seat model has some very interesting things to teach us, but I'm not sure what they are yet. A riding like Burnaby North-Seymour, which your projection had as Conservative blue all through the campaign, managed to be a complete Liberal runaway.

    I worried about the seat projection for the Liberals when I saw how massive a lead you were giving Joyce Murray in Quadra, and there might be something to that. Support may have bled from riding to riding once victory was assured in some, thus evening out the numbers and making the vote more efficient.

    No idea how to model that, though, and without a deeper look at the data it's just speculation at this point.

    ReplyDelete
    Replies
    1. Unless there is riding level polling I'd say its difficult if not impossible to model specific ridings...the swing model is doing what it is supposed to do and apply a broad brush over the region.

      Delete
    2. There may have been a 'favourite son' (grandson) effect in British Columbia.

      Justin Trudeau lived & worked in B.C. for several years of his life and his Mother's side of the family is from there (and we learned his Grandfather Sinclair was an MP/Minister there in the 1940's-1950's).

      Delete
  2. To add to that, I look forward to seeing what distribution your model produces when you plug the actual vote results into it, and then we can examine the discrepancies.

    Taken together, ridings like Edmonton Mill Woods and Calgary Confederation offer an interesting challenge.

    Also, I should acknowledge that my prediction of a record-low voter turnout was entirely incorrect.

    ReplyDelete
    Replies
    1. You might have underestimated Trudeau's ability to attract new voters, and Harper's ability to rally the anything but Harper vote to whatever party was best lined up to beat the Conservatives...in this case the Liberals. I know from your posts that you did not think the anything but Harper vote was going to hurt the conservatives, I mean directly it doesn't, but they did rally around one party in the end and sunk a good number of NDP candidates in the process.

      Delete
    2. The ABH trend was a bit stronger (and less precise) than I expected. It just moved votes from the NDP to the Liberals, without regard for which candidate had the better shot at winning. I think ABH took some seats away from the NDP and handed them to the CPC (in the BC interior, specifically).

      I mentioned Calgary Confederation because Éric's model had that as a reasonably confident Liberal win, but the Tories held it. In ridings where the Liberals had a chance, they underperformed.

      Only in the three unassailable Tory ridings in Calgary (Heritage, Midnapore, Shepard) did we see the Liberals exceed expectations. But in the ridings where the Liberals had even the slightest chance to win, it looks like there was an Anyone But Trudeau movement.

      Delete
    3. Ira, good analysis, but how did you think that it would be a record low turnout with there was such an extreme anti-Harper ferver throughout the country? I certainly have never seen anything like it. I met many people voting for the first time in a generation, and had conversations about the election almost every place I went. Are you in Alberta or Saskatchewan? When I was in Alberta last week, it felt like they were preparing for a funeral march - very much like the night Notley won. In the rest of the country, it was ridiculous. It mean the final vote totals were 68.5% in 2015, to 61.1% in 2011.

      Delete
    4. The polling stations were deserted in Edmonton, yes, but I made the prediction weeks before the election.

      1. I expected the CPC ads to do a better job of suppressing turnout.

      2. I expected the restrictions placed on Elections Canada's advertising would suppress voter turnout, particularly among first-time voters.

      3. I saw three party leaders, each with significant flaws. Mulcair's foreign policy was recklessly naive. Harper was widely seen to be pure evil. Trudeau didn't seem to be able to describe an economic policy in any detail.

      Admittedly, maybe the economic policy thing bothered me more than it would a typical voter. Look at Mulcair's cap and trade idea; there were no details about it in the NDP platform, when asked about it Mulcair said the details would be worked out later, but the platform confidently predicted it would generate $21 billion in revenue. How can you confidently predict revenue from a program whose characteristics you don't even know?

      Delete
    5. LOL, you're in Edmonton. Well that makes sense. I was in Alberta last week and I thought I was in the middle of a province wide wake. If you read the boards anywhere in the country you could see the ferver that Trudeau was producing. In Ontario it was intense. But what's very interesting, in Western Ontario, because of Wynne, it was very intense for the Conservatives. That's why I kept telling Eric that his Ontario projections were way off base, as Toronto and the urban 905 were going to go totally for the Liberals. Nobody in Quebec knew who would win, but most had a strong reason to vote. I got the same feeling when I was in BC. Out east the anti-Harper anger was most obvious, as it felt like a revolution. But you're right, in Alberta it felt like a morgue. That's why social media has become so important these days.

      Delete
  3. My gut says there was big 'anyone but Harper' effect going on that led to unusually strong strategic voting.

    ReplyDelete
    Replies
    1. Yes and no. For a lot of the BC interior ridings (Cariboo-Prince George, Kamloops, Vernon) the polls showed the NDP in first place. Everyone said to vote NDP to keep out the Conservatives. But after the dust settled, the Cons won, and the NDP finished 3rd. Even in my riding of North Okanagan - Shuswap (Vernon area), All three polls done near the end of the campaign showed the Liberals in 3rd.

      The less than honest Leadnow group only showed two of them to their supporters because they showed the NDP way ahead of the Liberals, and in first. The third poll showed the NDP only 2 points up on the Liberals in a fight for 2nd place. Cons won for only 39% of the vote. Had the vote not been split, the Cons would have lost all three of these ridings.

      Delete
    2. In the end the ABH translated not to voting for the candidate from the left with most votes but in voting for the LPC everywhere. In this sense Ira is correct that it lost some ridings for the combined opposition, but across the board it won a majority for the liberals. I believe we would have seen the same effect in favour of the NDP had Mulcair been the one ahead in the polls by election day.

      Delete
    3. I agree with Polstats' assessment entirely. Those voters were out to sink Harper and willing to sacrifice whatever party that was going to get in the way of that.

      Delete
  4. Can you please update the CBC Poll Tracker with the final results (Liberals 39.5%, Conservatives 31.9%, NDP 19.7%, BQ 4.7%, Green 3.5%, Other 0.8%)? I'm so very curious to see what the model predicts in that scenario!!!

    ReplyDelete
  5. Hey Eric,

    Going to listen to your 'casts in a moment but I wonder if you've looked at comparing the 1993 election to now, and then looking at the 1997 election and drawing some conclusions.

    Some interesting parallels to me.

    ReplyDelete
    Replies
    1. The Liberals managed to win all but 1 seat in Ontario in '93, and then all of them in '97. That makes those elections fairly odd, historically.

      Delete
  6. I ran a seat projection based on something Greg Lyle had mentioned. That the seat projection models, at least he felt, were underestimating Liberal seats, because they were based on swing from 2011. So I setup my projection model based on an average of 2006, 2008, and 2011. It spit out 172 seats for the Liberals. Still underestimated them but it called a majority, however narrow. Overall, called 61 ridings wrong, so not too bad. Aggregate absolute error for total party seats was was 42, or 8.4 per party.

    ReplyDelete
    Replies
    1. I've been thinking of seeing if something like that would do a better job.

      Delete
    2. True, seat projections are great in many ways, but obviously one of their limitations is that they use the numbers from the previous election as their base.

      As Greg Lyle correctly predicted, 2011 was unusual because the Liberal vote fell below 20% fro the first time in its history. That affected the seat projections.

      Delete
    3. Good point Greg. Maybe that is the problem.

      Delete
  7. I'm still hopeful that by the time of the next election, your site will be more accessible. In particular, I hope you will identify parties by more than just colour code, and present tables as real text, not image files. The current site is unnecessarily difficult or impossible to use for anyone with a visual disability, and fixing these issues would be easy.

    ReplyDelete
  8. Today's Globe discusses 22 narrowly won ridings. The 2011 CPC majority was won by 3200 votes for the weaker opposition party in 13 seats.

    I have not yet run the numbers for this election, but we need to consider NDP voters who switched to LPC to defeat Harper.

    A seat model needs to include the volatility of tight races, especially 3-way and 4-way contests where sub single percentage shifts can sway several seats.

    If Trudeau brings in IRV (instant runoff voting), I would expect better predictability in seat models if polling measures 2nd choices.

    ReplyDelete
    Replies
    1. What's an NDP voter? I mean consider these numbers...

      17.5
      18.2
      30.6
      19.7

      and these numbers...

      29
      37
      103
      44

      Those are the NDP vote and seat counts for the last 4 elections. I think it's reasonable to say that this years result looks a heck of a lot more like a standard NDP results then 2011's result and it would be more fair to say that the Liberal vote came home as opposed to the NDP vote moving anywhere.

      Delete
    2. The NDP had and has an English-Canada problem. They refuse to recognise the reality. The NDP's position against bicameralism for instance does not win them any votes. It may help solidify their base but, these people vote NDP in any case, opposition to bicameralism does nothing to excite Westerners or even most Ontarians and is anathema to many Quebeckers and French-Canadians in general. Dippers need to stop holding overly high opinions of themselves and listen to the people if they ever hope to be more than a third party.

      Delete
    3. The NDP lost because of their "English Canada problem" IMO.

      Although they have been removed back to third place, I think we should all remember that today the NDP is in a much stronger position than early Spring 2011 with seats in Quebec or even the Autumn of 1988. With a sizable Quebec caucus and representation from most regions of the country.

      Delete
    4. Calling the NDP results an "English Canada problem" is insultingly obtuse. The NDP don't do as well in the Maritimes (18%) or Alberta (11%) as these are the traditional strong points of the Liberals and Conservatives. They don't do well in Ontario (16.6%) because of Bob Rae. They had over 25% support in SK and BC.

      Delete
    5. Mapleson,

      You are "insultingly obtuse"-that is why I asked you some weeks ago to please refrain from commenting on my writing!

      Still blaming Bob Rae for the NDP's failure to be competitive in Ontario is a rather "insultingly obtuse" attitude toward the electorate not to mention completely unfair to Mr. Rae who left the premier's chair 20 years ago-Canada has held seven general elections since, he stepped down!

      Delete
    6. "Some weeks ago" was Monday. It's been a week now. Sorry, but you don't get to choose who reads and replies to what you post in a public forum.

      The Ghost of Bob Rae is still strong in many middle aged Ontarian minds, and that memory has been relayed to the next generation in political conversations. His return to politics as a federal Liberal stirred up the old talking points.

      There have been 6 federal elections and 5 Ontario elections since the end of Rae Days. However, that includes only two Ontario governments. When Bob Rae returned to politics in the 2006 election, the LPC dropped from 44.7% (75 seats) to 39.9% (54 seats) in Ontario. We can't break out what part of those numbers are due to Rae, but he still is relevant in Ontario, even if only as a boogieman.


      Likewise, The Ghost of Mike Harris is plaguing the OPC. The Harper Conservatives were much more centralistic and able to overcome the handicap.

      Delete
    7. Mapleson,

      One reason I asked you not to comment on my writings are the frequent mistakes and factual errors contained in your prose: " There have been 6 federal elections": How do you figure? I count 7 1997, 2000, 2004, 2006, 2008, 2011, 2015-7 Seven, VII-!; "and 5 Ontario elections". Wow! 1995, 1999, 2003, 2007, 2011, 2014!-WOW!- Addition not your strong suit eh!? 6 Ontario election six, 6, VI SIX!!!!!! Blaming Bob Rae for dropping the Grits seat count? Ridiculous, shameful and pathetic. Totally ignoring the reality of a unified Conservative party such a statement is vindictive and completely fanciful. Sadly, such outlandish statements demonstrate the NDP's inability to combat and correct what has long been successive poor election performances throughout English Canada. It is especially sad since, DIppers themselves are unwilling or perhaps unable to acknowledge the NDP's weakness throughout English Canada-that inability to acknowledge past mistakes all but, condemns the New Democratic Party to perennial also rans. They screwed up their one big chance, maybe in another 82 years they'll be competitive!!!

      I will also point out "centralistic" (sic) is not a word, and reminds me of another problem I often have with your writing: it is difficult to understand at times, sometimes through poor spelling and grammar but, often due to incorrect or garbled sentence structure.

      Delete
    8. Ghost, please review your history. Bob Rae was involved in the 1995 Ontario election. The discussion is the impact of the ghost of Bob Rae on the 2015 election, so it wasn't included in my list.

      The unified CPC does account for their higher seat count in Ontario, but their combined vote count didn't change much. However, the LPC in Ontario had a string of lower than "normal" vote counts until 2015, when it and total voter turnout rose up.

      The 2011 CPC win was a combination of many factors, first and foremost the combined base, but also LPC voter suppression (either by the CPC or unmotivated LPC supporters). Bob Rae has been a constant narrative in Ontario over the last two decades.

      FYI, http://www.thefreedictionary.com/centralistic

      Delete
    9. "There have been 6 federal elections and 5 Ontario elections since the end of Rae Days".

      No. There have been 6 Ontario elections since the end of "Rae Days" since, Rae Days as opposed to Rae's days ended before the 1995 election.

      Your writing is so sloppy it appears even you don't know what you have written!

      Delete
    10. I concede that pedantically "Rae Days" were applicable to only one year and that Bob Rae's final election and defeat as a member of the NDP occurred after that point. I'd assumed you were intelligent enough to understand the looser interpretation of "Rae Days" that since we were talking about the affect of Bob Rae on the NDP in Ontario since his departure from Ontario politics.

      Delete
    11. Mapleson,

      You are insultingly obtuse. Please respect my wishes and stop commenting on my writing. I will no longer respond to your asinine posts.

      I'm intelligent enough to understand English! You capitalised "Rae Days" thereby, writing something completely different than your latterly made up interpretation of your own writing! Your own writing imposes a stringent meaning upon Rae Days! You didn't write Rae's term, or time in office or Rae days, you were specific and assigned a unique historical context! You brought Bob Rae into this conversation not me so "we" were not talking about the affect (sic) of Bob Rae-Only Mapleson, was writing about the effect of Bob Rae on the NDP. I was writing about the NDP's English Canada problem-the party's inability to be competitive throughout most of English Canada.

      Regards,

      TGoPM

      Delete
    12. Ghost,

      If you want me to stop commenting, I suggest you stop writing. Otherwise, expect me to continue as I have since before you started using this pseudonym on 308. So long as Bob Rae comes up in a causal conversation about voting for the NDP within Ontario (provincially or federally), he is relavent to the issue.

      The NDP doesn't have an "English Canada" problem as they are very distinct regional strengths and weaknesses. Ignoring this is tantamount to ignoring the regional differences of Quebec.

      We need to wait for the release of breakdown by individual polling stations, but of the four NB districts with significant French-speaking populations, the NDP results were:
      Acadie--Bathurst 39.4% vs 50.7% LPC
      Beauséjour 15.1% vs 69.0% LPC
      Madawaska--Restigouche 25.9% vs 55.7% LPC
      Miramichi--Grand Lake 14.4% vs 47.3% LPC, 34.3% CPC

      Delete
  9. Éric, great work.

    Do you have any idea on how you'll approach modelling the next federal election? Are there any sources you can draw on for systems other than first past the post?

    It's obviously years too early to actually start building a new model since nobody (starting with Justin Trudeau) knows what the 2019 voting system will be. This question is about how you even approach the problem.

    Of course, the old model will be good for a few provincial campaigns yet.

    ReplyDelete
  10. aluchko I don't think any of the punditry or poll experts really know what the real level of this ABH is ?? Plus there is also an element of ABC in the results as well.

    ReplyDelete
  11. Eric,

    With all do respect I think you overly "massage" the data with the weighting system, incumbent bonus etc...By late September a number of polls showed the Liberals at 40%or better in Ontario and a majority government was within reach. You do a good job, however, I think an evaluation of your methodology is probably in order.

    ReplyDelete
    Replies
    1. The incumbent bonus is based on historical election results. The weighting system, if anything doesn't discount old polls fast enough. Just because some individual polls had the Liberals at 40% or better in Ontario, doesn't mean that those polls reflected reality until after the fact.

      The results were within the prediction, so the methodology is sound, if in need of some further tweaking.

      Delete
    2. No Mapleson,

      That is the point. The polls did reflect reality at the time of their commission-showing a shift was in progress. The seat project should have reflected the changing attitudes of Canadians. The projection was not within the Low-High seat range for the Liberals, Conservatives or NDP! it was in the min-max range. So with all due respect, I disagree with your last paragraph and would add that the projection was simply wrong. Whether this is due to the slow responsiveness of the methodology is a strong possibility, but, it seems, to me human error must also be a consideration,

      Delete
    3. Mapleson,

      Perhaps you did not notice my comment was not addressed to you.

      In fact the way polling works is that polls reflect the reality at the time of polling, polls are unable to reflect reality "after the fact"-what a bizarre comment. Unfortunately, Eric's projection was not within his low-high projection range for the Tories, Grits or Dippers only within the min-max range. I think any objective social scientist will always say: "improvements can be made". My understanding is that ideally seat projections should be in the low-High range depending on the party. Accordingly, 308 predicted: There was only an 8% chance the Tories would win 99 seats or less, a 20% chance the NDP would win below 51 seats, and a 12% chance the Grits would finish between 147-184 seats.

      Delete
    4. If you don't want anyone other than Eric to read and comment on your message, send an email.

      Without the actual election results, there is no proof that the polling was accurate to fall within the Margin of Error at a 95% confidence interval. The Min-Max range was added previously to fully cover the 95% confidence interval, while the Low-High range was a focused group within that.

      If the election turned out to be a Conservative Majority, the polls would have been said to be a miss. There accuracy can only be determined after their usefulness is terminated. I agree improvements can be made, as the call was on the outer edge of the 19 in 20 prediction. Eric is now looking at how a historical average riding result would have changed things (instead of just 2011 as a basis). I'd be interested in removing the historical voting from the equation by using riding characteristics to set a baseline (i.e. rural and rich areas are more conservative, poor areas are more NDP, ethnic ridings are more liberal).

      Other than ignoring the excluded 2.5% tails, your figures are correct on the probability of seat ranges. However, that doesn't make it a statistical miss. Just like the fact that rolling 7 on a pair of dice is most likely, any result 2 to 12 is possible. 3 to 11 would be within the 95% range, while 2 or 12 would be statically outliers. This election was 11 and not a 12.

      Delete
    5. Mapleson,

      Please refrain from commenting on my posts in future.

      Thank you,
      TGoPM

      Delete
    6. TGoPM,

      Please refrain from commenting in THE future.

      Thank you.

      Delete
  12. "the vote results should not have delivered a majority government to the Liberals."

    -> But Eric, since the Liberals hit nearly 40%, isn't that right in line with the mathematical threshold of where Majority Governments are normally formed these days?

    ReplyDelete
  13. Post mortem:

    1. The Notley bump at the Federal NDP level was short lived.

    2. Harper was really unpopular, around my circle of friends I couldn't find someone who liked it, but this seems to have held broadly across the country.

    3. Harper went so negative for so long that he forgot to create a legacy and reasons why to vote for him.

    4. As predicted the possibility of a close election brought people to the polls in higher numbers.

    5. Strategic voting played a role. This was clear from the outset. The question was who was going to benefit the LPC or the NDP.

    6. The Just not ready campaign was weak. In fact, this is 1 for 3 for negative ads: while there were negative ads against Stephane Dion, he pretty much did himself in regardless of the ads. They did work against Ignatieff.

    7. The supposed anti-PETrudeau effect was nowhere to be seen in Quebec and Alberta. PET is the premier most people love to hate, but when it comes to it people knew he was a forced to be reckoned and always had the best interests of the country at heart, even if sometimes leading to the wrong decisions.

    8. Separatism continues to look like yesterdays tired idea to a debate that was settled two decades ago (don't tell PKP though).

    9. Mulcair for whatever reason couldn't connect with the voters. In early September this election was Mulcair's to lose and lost it he did.

    10. People pay attention to the polls and adjust their opinions accordingly.

    11. A message of hope won over a message of fear or despair. There is no reason why this has to be so, but in this case people chose the candidate with the positive message over the one with the negative one.

    12. The last minute vote buying the day before the election was called was rather desperate. Was this a misguided effort to do some last minute vote buying or where the internal polls in the CPC campaign already reporting that people found nothing they liked about Harper and they were the CPC managers trying to do a last minute fix?

    13. The niqab issue drove away immigrant vote in the GTA. No I do not have data to back this up other than the ridings lost there, but this is how it seems to me.

    14. Careful you don't end up believing your own lies. Justin wasn't ready so need to worry about him right?

    ReplyDelete
    Replies
    1. 15. The 2011 result in QC was a max not a min. A year ago, I described this result as a perfect storm and unlikely to be repeated in 2015. I didn't expect a sharp fall either, but it always looked like the confluence of circumstances in 2011 was overly favorable to the NDP.

      Delete
    2. I like your analysis, and I would expand a little on why Just Not Ready didn't work. I think what did it in was another of Harper's strategies, the long campaign. He thought that the CPC's comparatively deep pockets would mean they could outlast the other parties. In fact, what it did was give Trudeau the opportunity to show that he was, in fact, ready. In a shorter campaign, JNR might have worked.

      Delete
    3. Just Not Ready was guaranteed to stop working at some point. The implication that comes with Just Not Ready is that he's Just Not Ready yet. But one day he will be.

      What this means is that as soon as Trudeau starts to look ready, the CPC have nothing else to say about him.

      They really should have kept the Not Ready ads to the pre-writ period, run a positive campaign to start, and then gone after his deficit plan as soon as he mentioned it.

      Delete
    4. Ira: we agree, running the "Just not ready" campaign for over a year was poor thinking.

      In terms of the deficit, I don't know if the attack would have worked. I grew up in a country that had a deep recession for many years. People personal cut expenditures and went into save-money-at-all-costs mode for quite a while. Eventually things started to get better and people started making once again long overdue purchases. At first they were all necessities, purchases that had to be made. After a year or so, the subject of holidays started coming up. Yes there were still other important purchases to be made, but no one had taken a holiday for over six years, and while fiscally and rationally no holidays might have been the safer course, people aren't machines and need downtime. So eventually, gingerly, people started going on short well-deserved three day holidays.

      I believe similar underlying reasons are behind the embracing of Trudeau's small deficit proposal. We just had six rough years, finally things are starting to look better, and it looks like the economy might need a small boost. So while people like you and Paul Ward rightfully obsess about the fiscal implications, the average person on the street welcomes the proposal.

      Delete
  14. Before jumping to any conclusions about seat predictions, we need to see what your model would have predicted using the real election results.

    Its also worth plugging this data into your chart for how often the model correctly predicts a seat. If it turns out that you were wrong in a lot of 50-60% seats but accurate in most 80-90% seats then it suggests your model isn't actually so far off.

    And on a side note, why does the model have a minimum percent of 50% for accurately guessing the winner? In some ridings the gap from first to third was quite small, suggesting an even smaller probability than 50% might be in order.

    ReplyDelete
  15. I am not so sure about the anything but Harper votes cost the NDP because that would suggest there was no fluidity in their voting intention in the first place meaning they were solid NDP votes but they were fluid and were not really NDP votes lost but rather undecided and there is a difference. Second I would be very interested in the results of vote splits - particularly in Quebec. With the rise in support to the Bloc I believe that this actually hurt the conservatives in some ridings as the split in the right wing vote may have allowed the Liberals to come up the middle, which likely they would not have been able to do without Harper waking up the separatist vote with the Niqab issue. Think he shot himself in the foot with that in Quebec.

    ReplyDelete
    Replies
    1. The CPC gained seats in Québec. I'd say whatever Harper said worked for him there.

      Delete
    2. I suspect a lot of the seats the CPC won in Quebec were the result of an incomplete shift in anti-Harper vote from the NDP to the BQ/Libs.

      Delete
    3. The BQ lost support this election, but it was more efficient, so less votes and more seats.

      Delete
    4. The seat gains in Quebec was caused by the exodus of NDP voters to other parties.

      Delete
    5. I'm inclined to agree with the Ghost. The seat changes in Quebec are largely the result of the collapse of the NDP.

      Remember, the 2011 election result was genuinely odd. When the Bloc collapsed, that vote had to go somewhere, and it wasn't going to go to Harper or Ignatieff, so it went to Layton. But those were never really the NDP's votes.

      That the Liberals managed to collect that vote this time is no less strange. The Liberals haven't won a majority of Québec's seats since the 1980 election.

      Delete
  16. I'm an observer of Canadian politics from across the river in the Detroit metro, as well as a number jockey in my past career, and I agree Eric did yeoman's work on this. I was absolutely addicted to Three Hundred Eight during the campaign. It would appear the ABH people were effective in tactical voting in the ridings where the LPC was the most viable alternative to the Conservatives. There might also have been some tactical voting in the opposite direction in the Windsor area, as the NDP were able to pick up Essex, which Eric had as a 71% probability of voting Conservative. I honestly don't know how one could model tactical voting to make the model a better predictor.

    ReplyDelete
  17. Always enjoy and appreciate your thoughtful commentary regarding the polls and poll taking.

    ReplyDelete
  18. Congratulations on the success of your site! I first came here for your analysis of the 2008 Quebec election and remember thinking how much effort you were going through for a blog that I thought only I and 6 other political nerds might read. I was as wrong about that as I was about the imminent swearing-in of Prime Minister Mulcair that I predicted two months ago.

    ReplyDelete
  19. Vote Together is claiming some success in a number of ridings. Their efforts may have skewed the model.

    http://www.votetogether.ca/pages/what_happened/

    ReplyDelete
  20. Excellent work! I'm very impressed by the work that you have done here and your ability to quantify your uncertainties. That first table which indicated the projected confidence and actual success rate of predicting those seats is a testament to your success. I'm guessing that there might be a little bit of tweaking to be done but as a scientist I applaud your excellent work and as a fan of the site I thank you for your efforts.

    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.