Monday, April 23, 2012

Final Alberta Projection: Wildrose victory

With mere hours to go before the polls open in Alberta, ThreeHundredEight.com is unable to project with any great degree of certainty what the outcome will be. While the polls were all lining up over the final days, the Sunday poll by Forum Research has indicated that the race will go down to the wire. But though it is possible that either Danielle Smith or Alison Redford will come out on top tonight, the seat projection model suggests that Wildrose will win, but miss out on a majority of seats by a whisker.

Assuming no MLAs cross the floor and that the largest party in the legislature forms the government, the model also suggests that Alison Redford's Progressive Conservatives will lead the Official Opposition and that Brian Mason's NDP will sit as the third party in the legislature with a large amount of influence.

However, there are a multitude of close races and a great deal of uncertainty in Monday night's outcome.

Wildrose is projected to take 38.4% of the votes, giving them 43 seats. That is only one short of a majority of the 87 ridings contested, a historic result for a party that won zero seats in the last election. The Progressive Conservatives are projected to take 35.8% of the vote and to win 39 seats, their worst result since 1967.

The New Democrats are projected to finish third with 11.4% support and to win five seats, their best result since 1989. The Liberals, meanwhile, are on track to win only 11.1% of ballots cast and are projected to be shut out of the legislature entirely for the first time in 30 years.

The Alberta Party is projected to win 2.2% of the vote, while the Evergreen Party (the re-incarnation of the de-registered Green Party of Alberta) is projected to win 0.6%. Neither party is running a full slate. Independents and other parties are expected to garner 0.4% support.

Closing gap in final week

Edmonton vote projections
While the election will be decided in Calgary and outside of the two main cities, where the Tories and Wildrose are fighting for every vote, recent shifts in the voting intentions of people in Edmonton could also play an enormous role in deciding the make-up of the province's next legislature.

After regaining support in the city in Week 3 of the campaign, the Tories dropped like a stone in Edmonton over the last few days before yesterday's uptick. They are  projected to win 36.3% of the vote and 19 seats. Wildrose has been relatively steady and is expected to win 27.3% of the vote and five of the seats, but the New Democrats have been making gains in the city. They are now projected to win 18.2% and five seats, putting them in play in one-quarter of Greater Edmonton's ridings. The Liberals, however, have fallen back.

Calgary vote projections
While Danielle Smith has been treading water in the capital, things are going badly in Calgary where Wildrose has been falling precipitously over the last 10 days. Though they still lead with 40.9% and 19 seats, that is a far cry from the almost majority support they enjoyed in the city for a brief period and the likelihood of a clean sweep that existed at the time. The Tories have benefited the most, closing to 36.6% and eight seats (saving Redford's skin in the process). The Liberals are just over the double-digit bar, a disaster for a party that had over 30% support in Cowtown in 2008.

Rural Alberta vote projections
In the rest of the province, the Tories have slowly increased their level of support over the last 10 days at the expense of Wildrose. Nevertheless, Wildrose leads with 44.5% and 19 seats to the PCs' 34.4% and 12 seats. Neither the New Democrats nor the Liberals are much of a factor outside the two main cities, and they have not been for the duration of the campaign.

The difference between a Wildrose minority and a majority will hinge upon whether Danielle Smith can reverse the negative momentum in Calgary and rural Alberta. For the Tories, they will need to capitalize on that negative momentum to squeak out a victory of some kind, which is still possible.

Where they stand and what is at stake

Each party has a lot riding on the results of Alberta's election. Can the Tories pull victory out of the jaws of defeat? Will Wildrose win a majority or a minority, or not at all? What influence will the NDP have in the next legislature? And what of the Liberals?

Click to zoom
Despite a few hiccups over the last days that may have cost the party an outright victory, Wildrose ran a strong campaign and may have simply been the right party, with the right leader, at the right time (others have highlighted Raj Sherman as the right leader for the Liberals but at the wrong time). As recently as Mar. 13 they were projected to win only 17 seats, itself a big leap for the young party.

Wildrose has the best shot at forming a majority government, with every poll giving them a lead over the PCs. They are projected to take between 36.4% and 40.4% of the vote. That range has loosened somewhat, which should come as no surprise as the penultimate polls of the campaign all had the party at the exact same level of support. Forum then mixed things up considerably.

Because of that potentially tightening race and some regional variations, Wildrose could win as few as 22 seats. This would put them in the role of the Official Opposition and would still represent a huge gain for the party, but compared to expectations it would be a disappointing result. Wildrose could also win as many as 62 seats, giving them the kind of landslide that is possible if a "kick the bums out" mentality takes over the electorate and boosts anti-PC turnout.

Wildrose will have the most success outside of Calgary and Edmonton, as they are projected to win between 40.5% and 48.5% of the vote and between 10 and 25 seats. They should also do quite well in Calgary with between 38.4% and 43.4% support and between nine and 22 seats, though the projection expects them to be at the higher end of that scale.

Edmonton should prove a more difficult nut to crack for Danielle Smith. Though she has been running neck-and-neck in a few polls, for the most part her party is trailing the Tories. They are projected to win between 22.3% and 32.3% of the vote in the provincial capital, giving them between three and 15 seats. A breakthrough of more than half-a-dozen seats would likely mean that Wildrose is going to win a landslide. They are projected, however, to end up at the lower end of the scale. Nevertheless, they should win enough seats in the provincial capital to give the city some representation in a government.

And that is what they are most likely to win. The odds favour a Wildrose government of some kind, but at such close margins it is a virtual coin flip. If the party greatly under-performs its polls, then Wildrose could end up as the Official Opposition. If they over-achieve, particularly Forum's results, they will win handily.

Click to zoom
In the week before the campaign was officially called, Alison Redford was in a strong position. Though she was only a few points ahead of Wildrose in the polls, it was expected that she would be able to handle the upstart party on the hustings. But on Mar. 29, she gave up the lead and since then has never regained it. Her only hope is that PC voters will turn out in great numbers and that supporters of the Liberals and NDP will opt for the devil they know.

There is less certainty about where the Tories will end up. Some polls have them in the low-30s, others in the mid-30s. This puts their likely range at between 33.3% and 38.3%. This makes it quite likely, but by no means certain, that they will finish behind Wildrose in the popular vote. But if it ends up being close, it can be expected that the PCs will have a more efficient vote.

This means that the Tories could win between 20 and 62 seats, like Wildrose a very wide range. If they win as few as 20, it would be an unmitigated disaster. If they win 62, it would be a second miracle result for the PCs. While those extremes are unlikely, the ingredients for a minority or even a majority do exist - but the Progressive Conservatives would need a lot to swing their way for that to happen.

Edmonton and Calgary will be key for the party. In Edmonton, they stand to win between 32.3% and 40.3% of the vote, giving them between nine and 23 seats. They are projected to be near the higher end of that range, though, meaning they need to win those extra swing seats to challenge Wildrose for government.

In Calgary, the Tories need to exploit the recent weakness in Wildrose's numbers. They are currently expected to suffer quite significantly, with between 31.1% and 42.1% of the vote and between five and 18 seats. Uncertainty is quite high in the city, but if the PCs can manage to have the votes swing their way in Calgary then they will have a real shot. If things go as projected, however, they have no chance as this is the only region of the province where they are at the lower end of their projected range.

In the rest of Alberta, the Tories stand at between 30.9% and 37.9% of the vote, enough to give them between six and 21 seats.

If they hold on to Edmonton and have luck in rural Alberta and especially in Calgary, then Alison Redford can remain as premier. But that is a best case scenario - the most likely result is that Redford will become the Leader of the Opposition.


Click to zoom
Prior to the campaign's start and the gains by Wildrose, Brian Mason did have a very good shot at taking that job. The New Democrats were looking good after Alison Redford became PC leader and the NDP moved into a three-way tie for second place. But now that the fight is between conservative blue and green, the NDP has its sights set on being the standard bearer of the non-conservative opposition.

Still, they are leaps and bounds behind even the second place Tories. With between 9.9% and 12.9% of the vote, they are almost certain to register one of their best results since 1993. They could win as many as eight seats, which would be a tremendous gain for the party, or as few as three seats, which still represents an increase.

Edmonton is their bread and butter, and they are projected to win between 13.7% and 22.7% of the vote and between three and seven seats. In all likelihood, any MLAs the NDP elects will come from this city. They have seen an increase in support of late, and whether they end up at the top or bottom of the range will depend on whether that vote turns out or not.

Elsewhere, the party is not likely to win any seats. Because of some gains in Calgary, they do have an outside shot at one seat, but it is a very outside shot. Rural Alberta is not expected to have any dots of NDP orange, though the party is doing relatively well with between 8.1% and 12.1% of the vote.

Because the Tories are too far ahead and the Liberals are falling, it is difficult to imagine the New Democrats finishing anywhere but third in the legislature. What is important for the NDP, however, is whether that legislature will be run by a minority or majority government.

Click to zoom
For the Liberals, meanwhile, the question will be whether the party will survive Monday night's result. The Liberals have had a rough time of it ever since the last election, but the campaign plunged their vote even further down when it became clear that the only realistic options for premier were Danielle Smith and Alison Redford. This has, according to the polls, sent a large number of Liberal voters Redford's way.

As a result, the Liberals are projected to place fourth in the vote with between 9.1% and 13.1%. They could win as many as three seats, but being shut out of the legislature is considered more likely. If any of their MLAs manage to be re-elected, it is probably going to happen in Edmonton, where the party has between 11% and 18% support.

In Calgary and rural Alberta, however, no Liberal is expected to win. Liberal support in Calgary sits at between 9.1% and 14.1% support and between 4.8% and 10.8% in the rest of Alberta. While surprises are always possible, the Liberals are not in a position to win any seats here.

And that means the Liberals are at risk of extinction, at least from the legislature. That is considered the most likely result, though sitting as the fourth party or even the third party is still possible. As Raj Sherman is likely to be one of those three potential Liberal victors, he may not exit the political scene in Alberta so quickly.

Click to zoom
The Alberta Party is not expected to be a major factor in tonight's vote, nor in any individual ridings. Though it is possible they could manage the unexpected, the polls do not point to any likelihood that the Alberta Party will hold any seats in the legislature. They were hamstrung by not being in the debates, and as the election has become about a choice between the PCs and Wildrose, the niche that the Alberta Party could have carved out for itself as a centrist alternative has been pinched out.

They have the highest potential in Edmonton, but the best shot for an Alberta Party victory is likely in West Yellowhead, simply because that is where their leader, Glenn Taylor, is a candidate.

An uncertain projection for an uncertain time

At this stage, it is usually possible to make a confident forecast. Until yesterday, everything pointed to a Wildrose majority government, but there were still so many intangibles that anything from a Wildrose landslide to a majority PC government was plausible. Now that a poll on the final day of the campaign showed a drastically narrowing gap, I have never felt less confident in a projection. The seat ranges, wide as they are, are a clear indication of why:
The last polls of the campaign had been relatively in agreement - certainly province-wide and to a large degree at the regional level. But the problem was that Wildrose and the Tories were running relatively close in a few areas, and with a few points swinging here or there the implications could include dozens of seats. If the gap is as narrow as two points, and was closing by a great degree in the final hours, it is very difficult to know where the parties will end up.

And this is not about hedging - there was a point in the campaign when the Wildrose and PC ranges did not overlap at all. Wildrose was poised to win a majority and only a majority, while the Tories had no hope of victory. And before Wildrose's surge, Redford was certain of winning another election for her party. The volatility and margin for error has simply increased to almost incomprehensible levels.

This is certainly not the kind of projection I want to make on Election Day. If I manage to get 86 out of 87 ridings correct, I could still potentially make the wrong call on the Wildrose minority. If I manage to get 84 out of 87 ridings correct, which would still be a stellar 97% accuracy rating, I could choose the wrong winner entirely! And I never like being in the position of saying a party will win no seats. But, I'm tied to what the model spits out.

This election campaign is also exceptional in several ways that make it very difficult to forecast. The biggest problem is posed by Wildrose, which took less than 7% in 2008 and did not run a full slate in that election. From there, they have gone to a lead in the polls. Can this sort of bandwagon be accurately projected?

The 2011 federal election in Quebec demonstrated that it can be. The seat projection model is capable of translating the NDP increase from 12.2% in 2008 to 42.9% in 2011 with a surprising degree of accuracy: it results in a total of 60 seats for the NDP (they actually won 59). But in 2008, the NDP managed to run a full slate of candidates in Quebec. Wildrose did not run a full slate last time, particularly in Edmonton. Where are their pockets of support likely to be in that city? As the capital and its surrounding region has roughly 1/3 of Alberta's seats, it is a large black hole for Wildrose.

The Progressive Conservatives, meanwhile, are seeing a wholesale change in the make-up of their supporters. Their voters have flocked to Wildrose and have been, in part, replaced by Liberals. That is a big demographic shift.

Final riding projections
Then there's the problem of the Liberals themselves, who are going from the main alternative to virtual fringe party status. Will their incumbents buck the regional trends? Could someone like David Swann keep enough of his votes in his Calgary riding to survive, while other Liberal candidates in the city do very poorly? Is there enough connection to voters' Liberal MLAs to ensure their survival? And what role will the Alberta Party play?

New boundaries could also mix things up, but that does not necessarily pose a problem for the projection model. Manitoba's boundaries were changed for the 2011 election, but nevertheless the model called 56 of the province's 57 ridings correctly.

What is different this time around is that I have added an adjustment to the polls to make up for the difference in voting intentions and voting behaviour, something that has become all the more apparent in recent years. This adjustment, which is based on how polls have been off of the result in other provincial and federal elections, assumes that the first and second parties in the legislature (in this case, the Liberals and the Tories) will be under-estimated in the polls while third, fourth, and fifth parties (in this case, Wildrose, the NDP, and the Alberta Party) will be over-estimated. Parties not in the legislature (in this case, Evergreen) are assumed to be greatly over-estimated. Had this adjustment been in place in 2011, ThreeHundredEight likely would have called the Conservative majority. It needs to be pointed out, however, that the adjustment is not based solely on that election's discrepancy.

What if that adjustment was removed, and the polls were assumed to be completely on the mark for this election? The result is a greater seat haul for Wildrose, all at the expense of the Tories, while the New Democrats still pick-up five seats. This result, however, is already encompassed by the projection's seat ranges. But it does show that Wildrose is well placed to win a majority. If Forum's poll is a complete outlier, then Wildrose will absolutely win 50 seats or more. If the momentum they measured is real and continues through to today, then the Progressive Conservatives could win that many seats.

An historic election

Alberta's election was never expected to be a very interesting one. While 2011 was jam-packed, 2012 appeared to be a very thin year for elections: Quebec maybe, and Alberta (yawn). Instead, observers from outside the province were treated to a political drama, while Albertans were offered a real choice for the first time in decades. And since the election will, one way or the other, result in the first female premier elected to govern a major province, Monday night stands to be historic.

But voters will be heading to the polling booths today without any strong idea of who will end up winning. While Wildrose is the favourite, the Tories could still manage to pull it off. And if the seat result is close, there is no telling what will happen in the legislature. On the other hand, a Wildrose landslide is still very much in the cards. Anything can happen. It is fitting that this surprising election campaign's voting day will hold more than a few surprises of its own.

88 comments:

  1. Your comments about defections, and Raj Sherman possibly still playing a role are intriguing.. Are you suggesting he (or the whole deminished Liberal caucus) might jump to either the PCs or the WR?

    ReplyDelete
    Replies
    1. Not necessarily, just that if it is some kind of close result that it is difficult to know what might happen. I was thinking primarily of right-wing PCs going over to Wildrose.

      I imagine the Liberals and NDP will try to squeeze whatever they can out of either party in a minority government, but that their MLAs will stay within the fold.

      Delete
    2. I feel that NDP would do what the federal NDP did: support Wildrose minority from outside government. Theyll support two confidence votes. Theres no chance in hell they support the PCs, knowing how wretched they are, even tho they may prefer them to WR. NDP will think that two years of Wildrose will scare Albertans, they can force an election on the third confidence vote and make a strong case to be made into a strong third party. They obviously wont make opposition, but I think they could build a lot of credibility in the left wing if they do that. Assuming, of course, the parasites dont move to BC and Saskatchewan.

      Delete
    3. I agree with your analysis of NDP and Liberal MLAs (assuming some are elected). Unless the Lib+PC= a majority there will be no coalition or policy agreement. The NDP will vote on a case by case basis.

      In regard to who will form government and possible defections, it can work both ways. If it is a PC minority, let's remember they remain government unless they resign or lose a confidence vote, some Wildrosers may be convinced to cross over especially if the PCs+Lib+NDP hold a majority. A cabinet post of patronage post can be very persuading. I also think that if WR does not win their party's cohesion will start to break down-what do they stand for other than toppling the PCs?

      Delete
    4. I also wanted to point out that if one party secures 43 seats they can in effect gain a majority by "appointing" an opposition MLA as Speaker. The Speaker always votes in the affirmative to continue debate. There are of course dangers with this method that I ahve not outlined.

      Delete
  2. I am surprised by this poll. It seems to be so much different than others, and everything else you have posted up to now.

    ReplyDelete
    Replies
    1. Indeed. But no one else was in the field yesterday, so we won't know until tonight if Forum captured something real or was off the mark.

      If there was going to be a dramatic swing, though, it might be expected to happen the day before people vote.

      We'll have to wait and see. I don't even have a gut feeling as to what will happen tonight.

      Delete
    2. A big final day swing makes sense to me, especially in this kind of election. It could, quite logically, represent a large chunk of the undecided vote making up their mind and breaking a particular way. In an election that, no matter how it turns out, is going to radically reshape the Alberta political map, it is to be expected that large numbers would be undecided going into the final days. This kind of election would definitely benefit with more final-day polling.

      Delete
  3. "unable to project without any great degree of certainty what the outcome will be"

    Nice use of a double negative. Suppose it never hurts to claim that your own model is able to project with a great deal of certainty what the outcome will be...

    ReplyDelete
    Replies
    1. Thanks, fixed. Give me some slack, I was up until 1:45 AM last night.

      Delete
  4. I'm inclined to believe you're putting a little too much stock into that final Forum poll. It seems to be the outlier, and not indicative of a larger trend.

    What truly causes me to question the reliability of that poll is that it shows PC gains at the expense of the WRP. I just can't see this being the case.

    The PCs may very well pick up votes, but it will be at the expense of the Libs and possibly the NDP.

    WRP support has been locked in for many days (and weeks) now. Anyone who has stuck with them this long will stick it out to election day.

    ReplyDelete
    Replies
    1. I'm tied to the rules of the model. Forum's poll is the newest set of data and has a large sample size, which gives it a large weight.

      Delete
    2. Nate, we don't really know whether or not it is an outlier because we don't have any other final-day polls to compare it to. The May 2011 federal election proved that public opinion can change dramatically in the final day or two of a campaign, particularly one as wild as this one has been. Just look at the final-weekend Nanos numbers for 2011, the dramatic CPC spike was there to see, but other pollsters weren't showing it because they weren't in the field, or if they were, they didn't provide a day-by-day breakdown of their results the way Nanos did.

      Delete
    3. Ouch. Hope you didn't wager your rent money on this ;)

      Delete
  5. One of the most fascinating and interesting elections in some time. A power struggle between the right and the far right, and nobody else can take advantage of it. If you beleive the hype it's a choice between the tired, old and socialistic PCs and the wild-eyed right radicals. And the west thinks Quebec politics are strange.

    ReplyDelete
  6. I think if a minority happens it will be more because of a stronger Liberal/NDP showing (10 seats between the two of them or more) with the Wildrose around 40 and the PCs in the low-mid 30s. It seems unlikely that the PCs will win a majority with only ~35% of the vote, but a last minute strategic vote and low turnout might get it done for them. Turnout will be higher but still under 50%. The most likely scenario is probably a Wildrose majority or at least a minority with them several seats ahead of the PCs, as the "change" theme is often hard to resist, though these parties have virtually no difference between them and no substantive change will take place.

    ReplyDelete
  7. While I am a huge fan of your work and this website, I think you may need to chuck your attempts at last-minute "adjustments". Last minute vote changes have benefited left wing parties, right wing parties, government parties and opposition parties. It's almost impossible to predict accurately who might benefit from a deviation of the raw data that comes from the pollsters. We should stick to the numbers as they are presented, rather than trying to predict how they will change.

    Particularly, I think it's a little strange that your model is treating the Wildrose Party as the 3rd party. They may be the 3rd party in the Assembly right now, but every single poll during the campaign has had them in first place. And given that its supporters are of the same fabric as the federal Conservatives, it is equally compelling to suggest that the polls might be under-estimating their support in the same way we saw last May federally.

    It's fun to play the prediction game and the what-if scenarios, but I think your site would have even more value if your model avoided trying to out-guess the pollsters. I'm very happy your chose to publish your "unadjusted" numbers - those ones look closer to what I expect to see tonight.

    Regardless, know that I will remain an avid reader of your work and a personal fan of yours. I once thought I was the only poll wonk in existence, or at least it sure seemed that way. It's nice to know that someone has really accomplished something through this website with the passion that we share.

    ReplyDelete
    Replies
    1. Thanks, that is very kind of you.

      I'm in a bit of a difficult spot, as people have often said I should go beyond the poll average. Now that I have, some are questioning it. I suppose I will never please everyone, but we'll see how this experiment turns out this time.

      Delete
  8. I've really enjoyed your work Eric.

    We political partisans, of course, want our side to win, but that being said, it's great to have a site free of spin and dedicated to something resembling a consistent methodology and analysis. That's really all one can ask for nowadays.

    ReplyDelete
  9. One factor I think that may (emphasis on may) be at play is similar to one that occurs in Quebec. In Quebec there's a trend for the PQ (and sovereignty) to poll a few points higher than actual results - it seems that 1-2% of people get to the ballot box and flip there. Pequistes call it the "ballot tax"

    I could see a similar result here based on two factors: (1) hesitancy over the wildrose and (2) the natural pull of voting for the same party you've voted for the past couple decades.

    There's no real way to account for that in this model (btw, Eric, I'm a huge fan of your work) but it may be a factor that's discussed tomorrow.

    ReplyDelete
    Replies
    1. "One factor I think that may (emphasis on may) be at play is similar to one that occurs in Quebec. In Quebec there's a trend for the PQ (and sovereignty) to poll a few points higher than actual results - it seems that 1-2% of people get to the ballot box and flip there."

      Voter turnout accounts for that

      Delete
    2. It's because people who want change are more motivated to vote. For that reason, a Wildrose supporter is more motivated than a PC voter. Similarly, PQ supporters are generally more motivated than Quebec Liberals because of that desire for change.

      Delete
    3. I think 1-2% of voters might be a little afraid to say Wildrose on the phone, especially in Calgary and Edmonton. But when theyre with their ballot, theyll gun it for Wildrose.

      Delete
    4. "PQ supporters are generally more motivated than Quebec Liberals because of that desire for change".

      If that is the case why does the PQ vote fall 1-2% at the ballot box?

      Delete
  10. Eric,
    I think you are underestimating the Wildrose in the rural areas and overestimating them in Edmonton.

    Thanks for awesome work on the polls. This site has been a lot of fun.
    Based on living in Calgary and constantly checking your site, I am going to predict....
    WR- 45
    PC - 40
    NDP - 2

    ReplyDelete
  11. WRP will need to win the popular vote by at least five points to have a good chance of winning a majority. If WRP wins the popular vote by three points or less they may not win government at all. WRP is likely to "waste" alot of their vote racking up big wins in rural seats. You might argue this is mitigated by the fact that rural seats have less population but they also have higher turnout percentages so those two factors will likely cancel each other out.

    ReplyDelete
  12. Eric, I just found your site and want to say thank you for your thoughtful, cogent and beautifully presented analysis.

    You have clearly put a lot of work into producing a very good product, and we all benefit from the information you provide.

    And I just want to add: ABW! Haha, guess we all can't help pulling for our desired outcome.

    ReplyDelete
  13. Prediction? Why not?

    PC 43
    WR 40
    NDP 3
    LIB 1

    ReplyDelete
  14. The only recent Forum poll I read about was this one.

    http://www.theglobeandmail.com/news/politics/wildrose-party-set-for-sweeping-majority-latest-poll-shows/article2410297/?from=sec431

    ReplyDelete
  15. I just realized I posted my prediction in the wrong comments section... oops lol.

    Here's mine:

    Wildrose 49 (+45)
    Progressive Conservatives 33 (-33)
    New Democrats 5 (+3)
    Liberals 0 (-8)
    Alberta 0 (-1)

    ReplyDelete
  16. Prediction

    WR 50
    PC 32
    NDP 5

    ReplyDelete
  17. I figure, I can't complain about Eric's predictions (not that I would, Eric system has it's virtues and limitations, and hefreely acknowledges the latter) unless I post my own (utterly unscientifc, and based solely on my instincts), so here goes:

    WR-48
    PC-36
    NDP-3

    For what it's worth, I think the "kick-the-bastards-out" sentiment will overwelm the "WR is scary" sentiment among true undecided voters who bother to vote.

    ReplyDelete
  18. I already was bold on another site and said Wildrose will get 50 or more seats, so Ill put that here. I dont care about the other parties, so thats all I have to say.

    Come on, Alberta!

    ReplyDelete
  19. I think the Forum poll may be accurate, and indicative of a last-minute reality-check of potential WRP voters. There are more than a few progressives voting PC just to stop WRP as well.

    Keep up the great work - much better analysis than any MSM.

    ReplyDelete
  20. Prediction

    Time for a change will be the dominant sentiment among late deciders, and, the small ray of false hope for the NDP in Edmonton will foil any organized tactical voting for the halfway accurately named "progessive conservatives."

    WR 58
    PC 25
    NDP 4

    ReplyDelete
  21. Wow, that last minute Forum poll has really thrown the cat among the pigeons.

    Won't know for another few hours whether it was the outlier, or whether Forum succeeded in capturing the last minute momentum.

    Fascinating election either way... my own projection, from a day ago, was quite different - WRP 55, PC 27, NDP 4, Lib 1 - let's see how things pan out!

    ReplyDelete
  22. I predict:
    Lib: 60
    PC: 27
    WR: 0
    NDP: 0

    Someone has to be the crazy outlier.
    And no, I'm not betting a dime on my "prediction".

    ReplyDelete
    Replies
    1. With a prediction like this it is clear that you weren't 'on the ground' for this election or for any other one in living memory.

      Delete
  23. Just heard Ian C, say on Evan's show that -don't be surprised if Lethbridge east elects an NDP mla. Where does he get those stats.

    ReplyDelete
  24. I predict:
    PC:40
    WR:38
    NDP6
    Lib:2
    AP:1

    ReplyDelete
  25. OK Eric

    By your numbers the PC's get 39 seats. The WR 43 and the others 5

    So add together 39 and 5 and we get 44. One more than the WR and it's not the new Govt !!

    ReplyDelete
  26. Hi Eric. Just cast my vote in Calgary and looking forward to watching the returns roll in this evening (although I suspect it will be a long evening ha ha! ). Just wanted to say I have really enjoyed following your updates and good work this past month. Your site is kind of the crack cocaine of polling sites for us political junkies! In any case wanted to thank you for the good work you have done and for caring about this election in our province. I am sure you will be up late like the rest of us watching how things shake out in this very special contest :) Cheers!

    Ken.

    ReplyDelete
  27. Mine is...

    NDP 62
    Libs 16
    WRP 9
    PC 0

    Wellllll...not really...

    But tonight is gonna be fun :-P

    ReplyDelete
  28. Kensingtonian has left a new comment on the post "Final Alberta Projection: Wildrose victory":

    I predict:
    PC:40
    WR:38
    NDP6
    Lib:2

    Looks a Hell of a lot more realistic to me !!
    AP:1

    ReplyDelete
  29. WR 53
    PC 27
    NDP 5
    Lib 2

    .... I am really looking forward to seeing the results.

    Wonder if Ontario will give credit to the Canadian leader in diversity of Alberta. Who will be on their 2nd female premier after tonight.

    I am thinking that 2 party leaders don't win their seats. The NDP is the only one I am sure will have a leader in the legislature.

    The rural will be pretty much swept WR. Edmonton is going to be really fun with lots of surprise splits. And I think if WR doesn't win about 2/3 of Calgary then they can consider this night a loss.


    Got my beer cracked, bring on the fun :)

    ReplyDelete
    Replies
    1. Who cares what Ontario thinks?? How is Ontario "credit" gonna benefit us in Alberta one way or another? Honestly...this evening Ontarian's should be more on pins and needles regarding the outcome of this election then we Albertans.

      Delete
  30. Lethbridge east may be capable of winning an NDP east. In the federal election the City of Lethbridge itself voted very highly for the NDP candidate but the overwhelming vote of the rural area sent it to Conservative.

    On a map of the Lethbridge East riding, Wild Rose Support starts right around #43-43. North Lethbridge is mostly poor and decaying.

    ReplyDelete
  31. Well I'll tell you bluntly this Alberta election is LESS important than the Ottawa-NY game tonight !!

    And unfortunately the Sens lost 3-2 !!

    So that means that no win Thursday night Canadian team is out of the Stanley Cup !!

    ReplyDelete
  32. To quote the legendary Kim Campell...oops...I mean Allison Redford...just after her teacher union bought leadership win..."Alberta politic's has finally caught up with Albertan's".I wonder if this evening when she is leading a hugely smaller PC party she will acknowledge that in actual fact it is Allison Redford and the PC party that has not caught up with Alberta politics! Afterwards will she do right thing? Tender her resignation?

    ReplyDelete
  33. Ohmigod!! The eastern bigwig CBC mucky muck's are in Calgary this evening! And on the National it was actually the top story!! How does my hair look? Does my ass look fat in these jeans? Christ....what if Chantal Herbert or Mansbridge not like me? Not sure what I will do...

    Bubba from the Blackfoot truckstop in Calgary

    ReplyDelete
    Replies
    1. Yes..they are. But unfortunately...we are still not good enough for them to interrupt regular broadcasting and televise coast to coast like they do for Ontario and Quebec elections. Instead the do "streaming" video lol!

      Delete
  34. I love how people are attempting to cast the Wild Rose party as the 'change' party.

    Welcome the new boss, same as the old boss.

    The political reality is that the PC party is the one that changed underneath the PCs; Alison Redford is clearly a Red Tory (no pun intended). The Wild Rose party is much closer to the Klein era PCs; all slash and burn, with a few reimbursement cheques tossed back to the electorate.

    If the WR party wins a minority, we'll see the same sort of situation as with the Harper minority of the last decade. The PCs will be forced to prop them up for fear of going back to the polls too soon (the NDP, of course, will never prop up the Wild Rose party).

    On the other hand, if the PCs win, you'll see them passing left-leaning legislation and the NDP will be the Queen-makers.

    ReplyDelete
  35. Like it or lump it sports fans but the results at 40 minutes after the polls close show Wild rose in a terminal DIVE !!

    Looks so far as a PC romp !!

    ReplyDelete
  36. It seems that as time moves on the Wildrose has LOST it !!

    At 52 minutes we have

    PC 56
    WR 21

    So can we
    say a decimation ??

    IMO we can !!

    ReplyDelete
    Replies
    1. huh ?

      WRA started this with 4 seats. On this trajectory they'll win a majority in 2016.

      Delete
  37. PC 60
    WRP 19
    Lib 4
    NDP 4

    At 11 PM Est !!!

    Can we say rout ?????

    ReplyDelete
  38. It'll be interesting to see if, given the real turnout (44% PC vs 34% wildrose instead of the other way around), Éric's model would have predicted the right result.

    It looks like it's mostly to blame on inaccurate polling/monster swing/incredible under-voting by WR supporters.

    ReplyDelete
  39. Well this looks like kind of a Wildrose wipeout. Hadn't expected that from all I've read of the Alberta goings-on over in Ontario.

    ReplyDelete
  40. What is notable here is just how misleading polls can be. There did appear to be dramatic momentum upwards in the PC numbers over the last few days. I guess the actual act of voting would appear to dwell upon the voters mind quite differently than talking to a nice chappie on the phone.

    I spend a lot of my time analysing sample data too and it never ceases to amaze me just how off the model predictions can be when sampling error is even a little wonky.

    I'm thinking Eric has the numbers right it was the sampling error and volatility of the popular vote that has made the result of this election very different than what many had supposed it would be.

    ReplyDelete
  41. Wow, was I ever as wrong as the pollsters. :/


    As a Conservative (a real one, not a CINO like Redford) I am very disappointed that Alberta has elected its first NDP government.

    As a Saskatchewanian tho. I am happy. Because we have alot of jobs that need filling, and even more as we try to resources to extract. I wonder if Redford will tinker with the royalty rate again and we will get to expand Saskatchewan even faster.

    ReplyDelete
    Replies
    1. "I am very disappointed that Alberta has elected its first NDP government."

      We few Alberta NDP beg to differ. The only difference I see between the PCs and the Wildrose is that the former are the devil you know, and the latter are the devil you don't want to know.

      Delete
  42. Clearly trying to predict the outcome of an election in a first past the post system using polling numbers is a futile exercise. The same thing happened on this site in the last election. Things can change so much on election day, I don't think any kind of algorithm can truly predict what will happen. I mean, the Wildrose party looks to have less seats than your low end estimate.

    ReplyDelete
  43. Wildrose Supporter (Former PC Supporter)24 April, 2012 00:08

    *head explodes*

    ReplyDelete
    Replies
    1. Hey there,

      I wondered how you'd handle the election results. I hope your head returns to normal soon.

      Redford 2012

      Delete
  44. I'm pleased to see sanity rule. I think the Harper government and the Redford government will work well together, although I expect Alberta to be a lot tougher about equalization.

    Smith's problem was the Yahoo's speaking out in the final week with their racist and gay basing views. Smith, unlike Harper embraced those statements, while Harper has always denied them. So far Harper is true to his word.

    Looks like Forum caught the trend, but not he magnitude of the trend, in their final poll.

    Good result!

    ReplyDelete
    Replies
    1. Smiths problem was the media twisting. A white candidate poorly wording a response in an interview conducted after the two other ethnic candidates. One of whom talked about "A white person could not represent the riding". The question was asked of all of them, about how your ethnicity will help you represent your constituents. The really scary part: "I think being white is an advantage". And the media went wiled on the white supremacist.

      Then there was the scary right wing gay basher. ... A pastor quoting the bible in writings and practice with the congregation and in church communications. Smith was badgered with questions daily about that all week. Noone thought to ask Redford her views on various religious underpinnings while she campaigned on the steps of churches, mosques and temples... all of which hold the same views as the pastor in question???

      Smiths problem was in the face of personal attacks by the left and the media like these, and like the tweet about her not having children and loads of others level for things like wondering aloud if the science over global warming is actually proven.... that she didn't go personal back just as deeply.

      If that is what the Campaign turned on... then Albertan's fully deserve Redford and the party that spends more per capita than any other government in Canada and most in the US. Rolling in revenue, and yet still hasn't balanced a budget in 4 years. Want to bet on the next 4??

      Delete
  45. This election signals an important shift in Albertan politics. Despite the PC Party continuing its 41 year dynasty, it is a much different PC Party now. A PC Party that would govern from the centre, instead of the centre-right.

    The Liberals garnered 26.4% of the vote in the 2008 election, now they are reduced to less than 10% of the vote. The vast majority of these voters transferred to the PC party.

    The PCs don't need to ponder to the far-right vote, as the Wild Rose caters to that segment of the electorate. The PCs will be comfortable picking up votes from the left side of the spectrum.

    ReplyDelete
  46. The biggest problem with your model is it assumes the accuracy of the polls it is based on. It was frequently stated in poll summaries, as it was in your predictions, that there were three geographic areas with different poll results. At the same time polls would normally poll approximately 1000 people across the province then state the overall accuracy was +/-2.5% which was clearly wrong for their assumptions. In order to get that degree of accuracy they needed to survey 1000 people in each of the unique voting populations, project the results onto the ridings in that area then add to get the total provincial result.

    I only saw one poll during the entire election that appeared to attempt that methodology. I think it was the Leger poll that predicted close to a tie.

    It is disappointing to me that the polling industry would use a methodology that assumes a homogeneous population across the province then in their analysis say that the population was not homogeneous but claim an accuracy based on a homogeneous population. It was even more disappointing that the media would take a headline from this analysis without having anyone critique the methodology.

    Perhaps someday things will change but it looks like that day is a long way away.

    ReplyDelete
  47. I have never seen pollsters agree so unanimously on numbers that turned out to be so, so wrong. Only the last day Forum poll pegged the PCs even within EIGHT points of the PC's final result. Abacus was off by THIRTEEN.

    ReplyDelete
  48. What about Eric, how do you explain this one?

    I am no fan of the Tories, and I regard Wild Rose as simply teabaggers, but you were way out.

    How about it?

    Arthur Cramer

    ReplyDelete
    Replies
    1. Hard to take the blame for this one, the polls were completely off. There appears to have been a shift in the final two days.

      My ranges did envision this sort of thing happening, and I was the only person seriously questioning whether Wildrose would win or not.

      But I can only be as good as the data I'm provided. How could I have guessed that the polls would be off by a combined 16 points for the PCs and Wildrose? I did assume that the PCs would be under-estimated and Wildrose over-estimated, something that I was criticized for during the campaign.

      Delete
  49. Results as of 12:55 a.m. MT

    PC 62 seats (Elected in 61; leading in 1)
    Wildrose 17 seats (Elected in 17)
    NDP 4 seats (Elected in 4)
    Liberals 4 seats (Elected in 4)

    I stand by my ROUT statement.

    ReplyDelete
    Replies
    1. Hi Peter

      How has your taste of victory in supporting a Centre-right party impacted your overall position on the political spectrum?

      I think that after firmly supporting Redford not supporting near identically policy governing Harper becomes somewhat of a dilemma for you and your cohort.

      I think that all of Canada really wants a centrist party that is fiscally responsible with a social conscience but not too influenced by the radical protests (from both the left and right) of the day.

      Delete
    2. I supported neither Redford or Smith.

      I think any right wing conservative has to be insane Unreason.

      I didn't vote for the obvious reason I don't live in Alberta.

      If I did I'd probably have voted NDP.

      The point I did support however is that given two bad options the PC's were the less bad.

      Delete
    3. I can't actually believe I agree with at least some of what BC Voice stated. Canadians increasingly want a more centrist government. And Harper and Redford are both governing in about the same place (center right) currently. The big difference is Redford is a centrist and is pulling her party slightly in that direction, which I believe the electorate has shown it wants. But Harper is one giant step right of where he currently is governing and is incrementally bringing us more right. And that is something Canadians are beginning to resist.

      Delete
    4. pink

      You are right in a way that things have moved "right" but let's look at this from a slightly different perspective?

      First Harper has dragged the national centre right by quite a distance. With this last budget he has set out to basically destroy all the social systems we have put in place. Expect stiff resistance there,

      Now Alberta PC's don't strike me as Harper Reforms !! That was Wildrose, not Redford's PC's !!

      Meanwhile we have the NDP nationally moving towards the centre. So it looks like the days of the "wacky left" are gone ??

      Redford has very smartly read the electorate mood and desires and planted her party there.

      I suspect that as things get harsher Harper will lose the majority and in fact by 2015 could well lose the Govt as well !!

      Delete
  50. Eric,

    I think it is not quit correct to claim that your prediction was as good as the data you had. More correct - as the data and your model was. The problem here is that your model doesn't attempt to extrapolate the polling data forward in time, with all the associated uncertainties. Years ago that would be meaningless, as there was not enough data for that, but these days, with frequent polls right until the election day, there is no excuse not to do it. Given the uncertain nature of systematic differences between different pollsters, I believe the best way would be to extrapolate the prediction in time to the election day separately for each pollster (and compute the corresponding extrapolation uncertainties), and then combine the predictions into one number, using the computed uncertainties, pollster biases etc., the way you do now. Only this way one could hope to reflect "momentum" or "trend" things in an election prediction model.

    Of course, one would have to make an assumption about the extrapolating function - a straight line would be the simplest. Given the function uncertainty, one should probably always report both the "static" prediction and the "trend" prediction, side by side.

    ReplyDelete
    Replies
    1. While I didn't use momentum, I did adjust the projection. Look at my projection compared to the "Poll Average Only" result.

      But even with momentum, there is no way I could've assumed such a huge swing with the data available. I could've gotten to a slim PC majority, though.

      Delete
    2. First, your proposed way of doing things would still not have caught the final spike in the PC vote, since only Forum was showing anything like enough movement to get the PCs to where the wound up. With this election, much like the federal election, what happened was a failure of the polls.

      As to whether the model failed, let's wait for for Eric's postmortem. It may well turn out that with the proper numbers, the model would have nailed the result.

      Delete
  51. An interesting election that is for sure. What is funny is the Wildrose party might have pushed Alberta (politically) to the left instead of the right due to taking away the far right wing of the PC party, leaving it to the 'Red Tory' side and pushing the Liberal & NDP voters to move to the PC's.

    Should be interesting to see what happens over the next 4 years and how it affects the next Alberta election.

    ReplyDelete
  52. 1 in 20 outlier24 April, 2012 11:09

    There should be a rider accompanying any Canadian poll release.

    This poll is for entertainment and politically campaigning purposes only.

    This poll is scientifically as accurate as examining goat entrails and making predictions on that.

    ReplyDelete
  53. Good morning Eric.

    You could have had your excuse article prepared for release this morning.

    This is my attempt at humor....

    Who could imagine that pollsters are just so far off or making things up.

    maybe you have to add 10 points to the party with the centre right agenda rather than 5 point as I have previously advocated.

    Maybe the people who vote centre-right (Harper, Redford) don't answer polls or just don't answer them.

    I think that the fact that robo calls (automated and call-centred) have a >95% rejection rate is just not a valid sampling technique.

    ReplyDelete
  54. "3. The undecideds decided

    Leger also asked a question that proved essential: have you decided? It found one-fifth of voters indeed were undecided in the final week of the campaign. The PCs believe many made their decision at the last moment. “I have to wonder if they walked in the polling station and went: ‘You know, my life is good, and the PCs have played a big part in that in government over the years’,” PC party president Bill Smith said. "

    http://www.theglobeandmail.com/news/politics/how-alberta-pcs-pulled-off-a-win/article2412042/

    ReplyDelete
  55. It'll be interesting to see what happens to WR now. They could spend the next five years proving themselves as an alternative government- Leech and Hunsperger will be brushed under the carpet for sure now they've lost- and try and destroy the PCs (whose house is, particularly with the survival of the Libs, still on very shaky foundations) next time. Or, given their spectacular seizure of crushing defeat from the jaws of crushing victory, the knives could be out immediately (Paul Hinman is gone, so God knows who the alternative to Smith would be), and the WRP could disintegrate into factions and disappear. We'll have to see.

    ReplyDelete
    Replies
    1. The WRA went from having 1 elected member to 17..... That is a much bigger gain than the Orange wave.

      Are you thinking that the NDP will disinegrate into factions and disappear?

      The knifes out for someone who brought the party from 1 to 17???

      PC's on shaky grounds because the Liberals survived?

      Wow... I wouldn't have ever thought that someone could spin this as a Liberal victory...

      I wonder if it will be Sherman or Swann who will step up after this provincal victory to lead the Federal Liberals.

      Delete
    2. If the WR's rise is so much more meteoric than the NDP's (which has existed for 51 years to WR's 4), then surely the WR is the more likely to be a flash in the pan. Especially considering that WR's a collection of social conservatives and libertarians united mostly by a desire to kick the PCs out, and that unlike the NDP, most people expected WR to get 40+ seats and probably kick the PCs out, not 17 with a PC majority mostly intact.

      And yes, the PCs are on shaky ground. Much of their support seems to be from Liberals who voted tactically to keep WR out, and people who didn't vote last time round. If WR can replace them on the Right next time round, and the Libs steal the centrist/centre-left ground to get the tactical voters back, then the PCs could drop right down next time.

      Delete
  56. One thing nobody here has talked about.

    The "Quality" of the WR candidates. If I'm not mistaken there were virtually no experienced politicians ?? Is the public stupid ??

    PC's of course has a virtually full slate of experienced house members and certainly more than enough to form a Cabinet.

    ReplyDelete
  57. Time to start "Ryan's Ye Olde Polling Shoppe." I 'll just throw darts at a dartboard and release the numbers I get as my results.

    The model did alright IMHO. The results were right on the edge of the range you gave, and I'd bet dollars to donuts that you're pretty much the only notable pundit to even suggest that such a result was possible.

    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.