Wednesday, April 27, 2011

NDP takes another seat, and why they might not win 50 more

Several national polls were added to the projection this morning. There's the EKOS poll from yesterday and the Angus-Reid and Nanos polls from today, along with numbers from Innovative and Oracle. The net result is one seat gain by the New Democrats, who are now tied with the Bloc Québécois at 43 apiece.

But the Angus-Reid poll puts the NDP closer to the Conservatives than they are the Liberals, and we've now seen over a half-dozen polls released in less than a week showing the New Democrats in second place. My projection still has them in third. There are a few reasons for this.

Most importantly, this is simply how the projection model is designed. It is meant to react slowly to new trends until they can be shown to be consistent and prolonged. With the NDP surge coming at the tail end of the campaign this might be coming too quickly and too late for my projection model to capture it completely. That's a limitation of the math - but is it also perhaps a reflection of what might actually happen on Monday night?

While the latest poll numbers would seem to strongly disagree with my projection, I actually believe my own numbers are closer to what the result will be in five days.

If the polls are right, we are witnessing a transformation of Canadian politics. That shouldn't be taken lightly. The potential results of 2011 have been compared to the 1993 election, but even that parallel might not be on the money.

The emergence of the Bloc Québécois transformed politics in Quebec, but considering what was going on at the time in the province it was an almost inevitable result. Supporters of the Parti Québécois were easily transferable over to the Bloc Québécois and support for sovereignty was running hot.

In the West, anger over how the region had been treated by the mainstream parties boosted the Reform Party to prominence. That it would be so effective was certainly a surprise, but that this sort of populist right-wing conservatism could find wide support in the western provinces is not outlandish. And, in the end, the Reform/Canadian Alliance was reborn into the Conservative Party, so it's almost like Western Canadians never left the conservative movement - the Progressive Conservative Party just left the conservative movement for a decade.

Now, the Liberal Party is to be crushed at the polls and be replaced by a party further to the left, but not for any particularly compelling reason besides the notion that Canadians like Jack Layton more than they do Michael Ignatieff. This hasn't been a campaign about the policies pursued by these two parties, and likeability may not be enough to get first-time New Democratic voters out to the polls. And this is a major problem.

The biggest boost the party has been getting in the country is in Quebec. In fact, more than half of the NDP's boost in support is located in Quebec. But according to their own people in the province, they don't have a local organization worth its salt in more than six or eight ridings. If they win elsewhere, it will be due to their rising support and not their own local efforts. A swell of sympathy for the party might get them 25% in a lot of ridings in Quebec, but in order to get them up and over the 30% mark that puts them in contention they will need more than a likable leader in the face of the well-oiled and experienced ground organizations of the Bloc, Liberals, and Conservatives.

And that is the difference between voting intentions and actual votes. Voting Conservative or Liberal or Bloc is nothing new for a lot of voters, so saying they intend to vote for those parties is not a stretch. Saying they will vote for the NDP, a party that they have likely never voted for in the past, is far less likely to hold true when they reach the ballot box.

The 1988 federal election is often cited as an example of when NDP support couldn't carry over to election night. In fact, the party dropped roughly seven points between the start of the election and voting day.

But I feel that the 2010 British election is a far better example. In that election, the Liberal Democrats were coasting in the polls, tied for second with Labour. While the Liberal Democrats are a centrist party, they do have a similar reputation in Britain as the New Democrats do in Canada: a third party with a likable leader, but a party lacking the gravitas of a government-in-waiting and a full slate of good candidates.

In the run up to the election, pundits had the Liberal Democrats making a historic breakthrough, taking them back to the days when they were a force in British politics. Nate Silver of FiveThirtyEight had the Liberal Democrats at 103 seats in his final projection. Other projectors had the Liberal Democrats winning a similar number of seats.

In the end, the Liberal Democrat vote toppled by about 4-5 points compared to their standings in the polls, and they only won 57 seats - which actually represented a loss for the party. The polls overestimated Liberal Democrat support and the party's weak organization failed against the stronger ones of the Conservatives and Labour. And, in the end, old voting habits were hard to break.

And that's why we shouldn't be so sure that these levels of support for the New Democrats will hold on Monday night. It isn't because the pollsters are wrong - they are tracking voting intentions after all. But correctly capturing the ability of these intentions to turn into votes is a very different thing for a party like the New Democrats, despite polls showing respondents "certain" to vote for them and unlikely to change their minds. Polls tracking the likelihood of Canadians actually voting often overestimate turnout by as much as a third.

UPDATE: A few commenters brought up the example of the ADQ's breakthrough in Quebec in 2007. It is a good counter-example, but I don't think the situation is the same. Mario Dumont was a fixture in Quebec politics for more than 13 years by the time of the 2007 election, and his party was following in the footsteps of the Conservatives' breakthrough in the province the year before. As Mario Dumont and Stephen Harper saw eye-to-eye on many issues, he seemed like a good premier for the province in the context of a Conservative federal government. And Quebecers who voted Tory in 2006 did not have to go very far on the spectrum to vote ADQ in 2007. The NDP is not in a similar situation in Quebec in 2011, though parralels in terms of the lack of organizational infrastructure and a strong team of candidates do exist.
Now that that's out of the way, let's get to the projection. The Conservatives are down 0.3 points to 37.2%, the Liberals are down 0.6 points to 25.8%, and the New Democrats are up 1.1 points to 22%. They have also captured a seat from the Bloc, and are now projected to win 43. The Conservatives and Liberals are unchanged at 146 and 75 seats, respectively.

Regionally, the growth of the NDP is clear. They are up big in the Prairies, Ontario, Quebec, and Atlantic Canada, and are also taking steps forward in British Columbia and Alberta.

The Conservatives are mostly dropping, as are the Liberals. But the Grits are spared losing a swathe of seats to the Tories - for now - as both parties are dropping at roughly the same rate where it matters.

The one seat to change hands is Drummond, formerly held by the Bloc. François Choquette of the NDP is now the projected winner.

Several other seats are very close to turning orange. In the West, the NDP is trailing by five points or less in Surrey North, Vancouver Island North, Nunavut, and Saskatoon - Rosetown - Biggar, all Tory seats.

They are not within that range in any seats in Ontario, but several in Quebec (Brossard - La Prairie, Laval - Les Iles, and Saint-Lambert) and Atlantic Canada (Central Nova, Dartmouth - Cole Harbour, South Shore - St. Margaret's, and St. John's South - Mount Pearl) are trending towards the New Democrats.

If they capture all of those seats, the NDP would be at 54 in total, with the Conservatives at 140, the Liberals at 71, and the Bloc at 42. So there is still plenty of room for growth for the NDP in the short term. Capturing another 50 or so, however, will be difficult.


  1. Thanks, Éric. I'm an undecided/NDP-leaner but I find your cautious approach so much more reassuring. I worry that people will be bitterly disappointed May 3rd with the outlandish 100-seats being indiscreetly thrown around. I know it's good for you to be accurate, but surely it's better for the party to be pleasantly surprised? Also worry about the party developing a bombastic complacency in this last week about sweeping the polls if the numbers are too inflated, and that's never good for anyone.

  2. I am quite concerned about aggregating multiple polls with different methodologies. Your prediction is just as valid as mine are, by the nature of the beast, but I am concerned about your methodology.

  3. Looking at this morning's Nanos poll, it shows a massive 21-point lead for the Conservatives in Ontario, with a split of 46.9-25.7-21.0 (Con, Lib, NDP). If Nanos is correct in that the Conservatives are in the mid-40s and the Liberals are plummeting, how would that change the result? In other words, how many more seats would the Conservatives win if they were five points above what is in the projection currently and the Liberals were five points below where the projection puts them currently, in Ontario? Am I correct to think that would flip at least 9 seats over to the Conservatives?

  4. Great work as always, Éric.

    Do you have any plans to adjust your Renfrew-Nipissing-Pembroke projection to account for Hec Clouthier, who, whilst still an underdog, will surely garner more than 0.7% of the vote?


  5. I'm just curious what a projection would look like using either the Ekos or the AR polls. The projections that are used may end up being right but these polls all seem to be saying something similar right now

  6. I agree with your caution, Eric. I am and NDP partisan, but I think the party is better to have the chance to be pleasantly surprised by relying on cautious projections like yours rather than risk serious disappointment by relying on outlandish projections like Frank Graves'.

  7. During The Insiders on The National last night they pointed out that in a surge like this even the latest polls may be underestimating the NDP's growth because they are two or three days old. Averaging with polls that are even older makes you numbers even more out of date.

    Perhaps instead you should add a factor reducing the vote in ridings for parties has moved from a few votes to many.

  8. Did the Bloc and Reform have excellent ground operations in 1993? Nope. When there is this much momentum it makes ground operations redundant.

    Ps. The NDP wrote the book on GOTV in Canada.

  9. I'm in total agreement with you. I used to vote NDP and seeing them win big would have a certain satisfaction for me even though I am not voting NDP. But 100 seats? It's a fantasy. Even becoming the official opposition is I think far-fetched at this point. If the Conservatives win a majority and the Liberals lose seats, there may be a huge readjustment within the Liberal Party that makes them vulnerable to the NDP in the long run. The problem is, this surge for the NDP is based almost entirely on Layton. Can he realistically stick around for another election if the Conservatives win a majority?

    My guess is, if there's a Conservative majority, the Liberals will have four years to rebuild properly and with the right leader come back the next time around. In four years, Canadians will be sick of the Conservatives, as they would be of any party in power for nine years.

  10. This is reminding me a lot of Ontario in 1990 when the NDP surged ahead in the final polls. Everyone kept saying "the NDP will never be able to turn votes into seats", "they have no ground game in most ridings", "so and so (insert name of your favourite Liberal or Tory hack) has been there for 20 years and is unbeatable" etc...well the votes were counted and the NDP went from 19 seats at dissolution to 74 seats - and swept ridings where they had no active campaign at all! We should also not forget the Quebec election of 1976 when the PQ went from 6 seats to 71 in one fell swoop!

    It may be true that the NDP has no organization in Quebec outside of 6-8 ridings. I have news for you - the Liberals are organizationally non-existent in Quebec outside of the west end of Montreal and the Tories are organizationally non-existent outside of about a dozen seats they are competitive in. The BQ/PQ supposedly has an organization - but when the ADQ swept Quebec in 2007 it wasn't enough to prevent some supposedly unloseable PQ seats in Monteregie from going ADQ. Ground game only matters in a close race.

  11. Interesting analysis Eric but I am not certain I am fully on board with it. I understand your reasons and the logic but, with the current slew of polls, I am afraid you could be significantly off come election night.

    The examples you cite to support your decision are indeed correct but you are omitting counterexamples. Situations like the ADQ coming out of nowhere to capture a whack of seats in Quebec, Japan's LDP losing to the opposition they used to ridicule or PC in Canada being reduced to 4 seats under Kim Campbell.

    Furthermore while your theory that people's voting intentions may well change and go back to the tried and true choices has some gravitas, I am thinking that with just 5 days left, there is very little opportunity for "buyer remorse" and it is more likely that the voters will indeed vote their current preference.

    For the record, I have no particular political allegiance, but feel that the trends coming out are just too strong to ignore. It's your decision of course but, if these polls continue with the same trend today and tomorrow, I would seriously consider revising the model. As they say if it walks like a duck and quacks like a duck, it probably is a duck! :)

  12. Eric, I really appreciate the experiment you are running here. Early in the campaign, I found it a very good way of sorting out the wheat from the chaff amongst contradictory polling results. But while your methodology may be good for improving the signal-to-noise ration in a relatively static or slow moving situation, it may be too cautious in a wave situation like the current one. With three national polls in the last 24 hours showing the NDP between 27% and 30%, there is no way that current voter intentions would give them 22% of the vote. You may be right that current intentions may not turn into votes in the ballot box, but shouldn't your model show what would happen if current intentions - properly measured and weighted - did turn into actual votes?

  13. The fact of the matter is it is too late to change anything. I have to stick with the methodology, and this unexpected campaign might throw me off. That's the numbers and I have to live with them.

    My explanation for why I think they might end up being right is my gut talking. There are counter-examples, sure, but either my examples or the counter-examples will prove right on Monday night. I think my examples will prove right, but I'm very aware that I might end up being very wrong.

    But there are still five days left, so a lot of new data will be added to the model before election day. What I have now is not what I will have in my final projection Sunday.

  14. Ryan,

    The Bloc did - it's called the Pequistes. They're the same ones who helped the Mulroney PCs.

    Reform is slightly similar to your point, since they built up over an election and inbetween, as the NDP have. However, they also had a lot of help with disaffected jumpers from the PCs and a lot of the provincials onside as well. The NDP don't have this in either case.

    As for the NDP "writing the book," well, sure, if it takes you 6 elections to do it, they sure wrote that book.

  15. One thing that I have not found in your analyses, is the number of missing liberlas, if they were to come back into play.
    Do you account for these

  16. Any idea what would happen if the Green Party went poof?

    Also, this just goes to show what I noticed years ago here in the US: third parties dont know how to win elections. You need the ground infrastructure in order to win higher and higher offices. Similarly, the NDP was caught completely off guard in Quebec and cant do much. On the other hand, I imagine theyre limited on funds.

  17. Any comments on the poll that puts the Green Party ahead of the Conservatives in Saanich-Gulf Islands?

  18. A note about the 2007 example of the ADQ.

    Mario Dumont and the ADQ had been on the Quebec political scene for more than 13 years in 2007. Jack Layton has been leader for seven years, but no one in Quebec really paid much attention before 2006.

    Secondly, the ADQ's surprise performance in 2007 was following the Conservatives' surprise performance in 2006. Dumont looked like the kind of person to deal with Harper since they saw eye-to-eye. And the Tories had done it in 2006 so voting for the ADQ in 2007 was not a stretch.

    There is no similar parallel for the NDP in Quebec in this election.

    The ADQ in 2007 is a good example of what can be done without a ground game or more than a handful of quality candidates, but the context was favourable to them then in a way it isn't for the NDP now.

    I'll think I'll add a small note in the vein to my post.

  19. Hi,

    The "ground war" doesn't matter in a last minute swing. I worked with the LibDems in a constituency where they were a close 2nd in the 2005 UK election. Despite the fact this was a targeted constituency with a massive local operation, it nonetheless felt the same national swing as every other riding, and they were again a close 2nd in the 2010 election.

  20. With all due respect,I think your model suffers
    from some constipation.The trend in Ontario is
    increasingly in favour of conservatives sailing
    to some 70 seats.It becomes hard not to see a
    looming conservative majority as all the hype
    about the NDP surge will surely lead conservative Libs in voting CPC to protect against socialism.Sagging liberal fortunes in
    the atlantic also spell important losses in seats for them.Would you agree to make projections on 1)The Nanos tracking polls and
    2)The Ekos tracking polls?Poll history is
    probably not very pertinent in a context of
    the wild swings we are witnessing these days.
    I read you every day!Thanks for your service.

  21. I think I said earlier an another thread that this kind of "surge" or dramatic change throws all the polls and polling out the window.

    Thanks Eric for proving me right.

    Short of things settling down over next weekend "??" this is going to be a major paradigm shift.

    Just as an aside I wonder just how much of this change is due to ABC ??

  22. The points you make about the importance of local organizations and the example of Britain's Liberal Democrats are valid, but let's not discount the fact that the public can swing strongly in certain instances. The Ontario NDP in 1990 didn't have a good organization on the ground either outside of its strongholds, and yet a provincial swing to 37% was enough to push that party into a majority, with a number of rural seats.

    The mentality of some Bloc voters may be, "okay, Harper: you say no coalition is valid with Bloc support? We'll vote NDP then. What do you say to that?"

  23. Ontario Dipper

    One thing for sure, the imlosion of the Liberal vote in the west, and the BQ in Quebec changes a lot of things. Also the Conservatives are down about 10% in Quebec in a lot of polls, and they are vulnerable there, as well as in the west where vote splitting between the Libs and NDP looks like it is gravtating towards the NDP in a big way that will threated another dozen Conservative seats out west.

    Ontario is where the Con majority will live or die, and the Orange Crush appears to have nhavigated the Ottawa river at long last. I'm pretty sure Ed Broadbent never got more than 24% in Onraio, and the latest polls (other than Nanos, whose numbers are skewed by taking an Easter two day break which makes many of their numbers 5 days old)are showing the NDP over 25% in Ontario. Many 416 ridings (Davenport, High Park, Beaches East York, and others are) now leaning NDP, and even the Conservatives are vulnerable in Essex, Oshawa, and Kenora.

  24. The Lib Dem example looks like a good example on the surface... but once you dig deeper, it's a pretty poor comparison to the current NDP surge.

    First, Clegg ran an extremely poor campaign the last week of the UK election up to the polls, with latter polls beginning to reflect this. So far, Layton has not done this, and in fact, his momentum has continued to increase.

    This is looking more like NDP win in Ontario in 1990, Nova Scotia in 2009, and the ADQ in Quebec. These elections are how Canadian surges operate.

  25. Éric,

    The ADQ's "surprise performance" in 2007 had a lot to do with the impopularity of Jean Charest and the PQ's choice of leader (André Boisclair -- remember him?). Mario Dumont was similar to Layton in that he appeared dynamic and clear-spoken and yet was perceived as an outsider, not tainted by association with "old parties". (Perceived that way -- in reality he was a former Liberal who campaigned on the Yes side in 1995...)

    If anything, I suspect Quebecers are more willing to cast a "let's try the other guy" type of vote in federal elections than in Québec elections.

    All of this to say, we have no idea what's going to happen Monday. Your caution may be warranted, and you should certainly stick to your model -- why else have this website? But an NDP vote anywhere from 20% to 35% (across Canada) seems possible.

  26. I think it's time to look at the behaviour of the campaigns to see which polls may be closest to what they are seeing internally. The Conservative campaign has moved into do no harm mode almost a non campaign, which means they are ok with where they are, aka majority. The Liberals are flailing around like a fish out of water. The NDP have Layton on every news show possible in Quebec, but they themselves released the story that they only have an organization in 8 ridings in Quebec. Tome this means, the Conservative internal polls see a majority and are mostly ok with the NDP rise, the Libs are in deep trouble even in Ontario, and the NDP know they won't get as many seats in Quebec as some outrageous pollster suggested. I would say a max of 15.

  27. Éric, I must say that I disagree completely with your interpretation of the ADQ rise in the 2007 Québec elections. They had nothing at all to do with the Conservatives and the 2006 elections and people certainly didn't vote for the ADQ to have people who saw "eye-to-eye" with the Conservatives.

    What happened was that the ADQ was able to come through the middle between an unpopular incumbent PM with very low satisfaction levels amongst francophones (Charest) and a new unpopular leader of the Parti Québécois. Also, Boisclair was gay, and I have heard of people who voted ADQ just because they didn't want to vote for an homosexual, maybe they weren't that many, but just 5% of homophobes can turn a PQ seat to an ADQ seat in certain cases. There is also a great desire for change in Québec and a bit of exasperation with the two major parties and the old debates, and the ADQ tapped that desire enough to have a breakthrough without an organization on the ground.

    Note though that the ADQ's performance as official opposition ended up being so poor that they lost almost all their gains in the next election, leading to Dumont leaving the party and political life.

    As to the comparison with the NDP, there are some similar points. Like the PQ in 2007, the Liberal party has an unpopular leader that doesn't inspire people. Like the ADQ in 2007, the NDP has a popular leader and has successfully labeled itself as the party of change. Like Charest in 2007, Harper doesn't really offer a vision to people to justify voting for his party, merely defending his record, and he is the one people most see as a competent Prime Minister. Again like Charest, Harper relies on a strong core of voters who seem strongly tied to his party, but has also a strong core of people who oppose him, so that he has little place to grow (see the Conservatives' pitiful "second choice" results in the EKOS polls).

    However, there are also differences. I do not feel the same desire for change in the Rest of Canada that I feel in Québec. The distaste for the two main parties isn't as pronounced in Canada as it was in Québec in 2007. However, in Québec the dynamic is really present, none of the traditional party leaders (Duceppe, Ignatieff and Harper) inspire, Harper is downright hated by a vast number of people and disliked by even more people than that. There is the recipe for an organization-less breakthrough in Québec for the NDP.

  28. Here's a link to the riding poll for Saanich-Gulf Islands

  29. Anonymous said...

    OK so lets just look at Nanos for a second, a fairly good pollster that no one really accuses of having a bias one way or another.

    Nanos has the Conservatives at 47% from 39% in 2008 in Ontario and the Liberals at 26% from 34% in 2008. These numbers have been consistent for several days.

    That's a Conservative Majority folks.

  30. For what it's worth, I'm a Green/Liberal voter who'll be voting NDP this election simply because of the poll numbers (even though I disagree with NDP policies and dislike my NDP incumbent).

    I believe that what this country really needs is electoral reform, so that no one ever has to hold their nose and vote "strategically" again. If NDP has any presence of mind, then the first thing they will do after becoming the second party is propose electoral reform. They've been screwed by the ugliness of our backwards first-past-the-post for too long.

  31. Well Eric I both agree with your note of caution and yet agree with DL that there have been a variety of major shakeups in Canadian elections that looked similar. I don't know what to expect.

    Yet... I really feel that the split in Ontario could determine more than any NDP breakthrough. If the CPC gets a majority then whomever gets elected anywhere else won't matter a whit.

  32. Great blog, just wondering if (without having looked deeply into pollster methodology) if there needs to be an assessment of the poll numbers considering that they use land lines and would therefore skew toward an older demographic in their respondents? Given that the cell phone users tend to be younger, and that younger voters tend to vote NDP more than other parties, there might be a slight suppression of voting intention. Of course, the same youth tend to vote less, so maybe it cancels itself out.

  33. Most likely you have already seen it Éric, but Le Droit released a local poll for the riding of Hull-Aylmer showing the NDP well ahead

    NDP : 42%
    Lib : 29%
    Bloc : 13%
    CPC : 11%
    Green : 2%

  34. Keep up the good work, Eric. As election day approaches there are bound to be endless posts from many nervous partisans, but stick to your guns and keep giving us the model's output.

    I also enjoy your asides which detail why particular outputs may or may not be accurate; on the whole, your model is the one that I respect the most due in no small part to your transparency and rigour. Keep it up!

  35. James Bow,

    The NDP were part of a governing coalition in Ontario before 1990, and had an organization including massive union backing, and the public sector, in every riding in the province. They were highly organized.

    This is VERY different than having no history and 8 offices in an entire province. You can't run a winning campaign from 8 offices against the fiercest political machines in the country. Some of their candidates don't even live in the province. One is a waitress in Ottawa, another a 71 year old that can't be found.

    It's not the same thing at all. It's not the same as the ADQ either, who were also as unorganized as they were, light years ahead of the NDP in organization in Quebec.

    I'll give them 2 seats for every office, which is a lot. More than that is fantasy. They may have a very strong showing in many ridings. Winning is another matter.

  36. James Bow,
    The ADQ was also aided by the Conservative machine around Quebec City.

  37. I'd just like to point out to everyone that I believe the objective on this site is to be as accurate as possible on the last day of projections - not to be as accurate as possible on every intermediate date.

    As such, I think the comments that the projection is too cautious, etc. are misplaced. Eric cautions everyone that he's showing us snapshots of a model as it progresses from beginning to end. As a parallel, complaining about the results you are seeing now is like complaining about a movie's ending when you've only seen the first 45 minutes.

    I'll save my objections / comments on the methodology until the election. And until then - thanks Eric for your insightful posts.

  38. Your (and CBC's) comparison of the NDP surge to the Lib Dems of the UK is incorrect. British pundits universally agree that Clegg ran an awful campaign in the final week, and I don't see Layton having done anything of the sort, nor really even capable of making a flaw critical enough to turn his momentum into the other direction.

    His Quebec policy, quite frankly, has no traction outside the province. Canadians are generally supportive of the notion to finally include Quebec in the Constitution; it is only a toxic issue for the politicians. And any alarms that might have gone off thanks to Harper/Ignatieff fearmongering would have been pacified by Layton's statement that Constitutional talks would not be an immediate priority. As for inside La Belle Province, it is clear that the soft nationalists don't have a better party to turn to than the NDP.

    But yes. Layton =/= Clegg. False dichotomy, although I would agree with your call of caution against NDP supporters letting their ego get ahead of themselves.

  39. Anonymous at 11:05:

    I'd agree that, if Nanos' poll for Ontario was spot-on, there would likely be a Conservative majority. But that's a big IF. You picked one poll only, even if Nanos isn't biased it still wouldn't mean that it is necessarily right.

    There are other polls.

    Angus Reid was the most accurate pollster in the 2008 election (and in the 2007 and 2008 Québec elections). It claims 37% for the Conservatives in Ontario, 30% for the Liberals and 27% for the NDP.

    Ekos, the second most accurate pollster in the 2008 election, says Ontario is 37% Conservative, 32% Liberal and 23% NDP.

    I'm not saying Angus and Ekos are necessarily right, but the truth is likely in the middle. You can't just pick the poll that you like best... unless you just want to do some political spin, of course. But this isn't the place for spin, this is the place for political analysis.

  40. It may be unrealistic to suggest the NDP will win a hundred seats, but there are many examples, the 1993 election being the most cited, where the near-impossible happened.

    Eric, it would be great to be able to compare a one-poll projection from you using the numbers that are showing 100 seats for the NDP. If your model says something drastically different, we know we're in messy territory. If however, your model would give them upwards of 80 seats, something very real is happening here.

  41. Anonymous 11:31,

    Pollsters don't just call land lines, they also call cell phones.

  42. "And Quebecers who voted Tory in 2006 did not have to go very far on the spectrum to vote ADQ in 2007."

    You mean like Quebecers who have previously voted for the left-leaning Bloc and are now voting for the similar NDP?

  43. @Anonymous (11:05)

    There are also several other pollsters, who are also considered "fairly good" and never really accused of having a bias one way or another, who have the Conservatives in the low 40s or high 30s.

    As Anon (11:49) pointed out, the two pollsters who came closest in the last election both have the Conservatives at 37% in Ontario (a drop of 2% from their popular vote in Ontario in 2008). They're both at the opposite end of the spread from Nanos right now, but just as valid and just as likely to be true.

    It's probably best to consider them the current floor and ceiling - the 47% Ontario ceiling puts them into majority territory (as it has all election), and the 37% Ontario floor has them losing a few of the seats they had in the province. The truth could be anywhere between them, and could always shift between now and election day.

  44. I didn't pick the poll I like. I actually think Ipsos is closest. I picked the pollster in the middle,who also has a VERY good track record.

    If Angus Reid and Ekos are polling accurately, we would see a very different campaign from the Conservatives this week. These two are also polling at the bottom of the pack for the Conservatives, so I doubt they are correct. I also predict their Tory numbers will change before the vote, given the other pollsters numbers.

    Nanos was closest in the previous election to the last one, within .1% of all parties. It is polling in the middle of the pack this time. That's about as neutral as we can get.

    Again, the Conservatives don't look too worried this week, maybe they should be, but I think they have some other information.

  45. BT, The Conservatives have gone into shut down this campaign mode and are talking post election now in their events. They wouldn't be doing that if their internals showed them losing seats in Ontario.

  46. About the ADQ. The PQ's coalition had been between urban separatist progressives and leftists along with rural traditionalists and many who can be described as nativists, even xenophobes. Economically they responded to populism. The PQ depended a great deal on those rural ridings with that component.

    When they selected Boisclair they did it in part because of embarassment regarding what was seen as racist elements in the PQ, particularly charges of anti-Semitism. This was led by Lucien Bouchard who was Boisclair's patron and who resigned over the Yves Michaud affair.

    Boisclair essentially alienated these people in part because he was gay but also because he was so obviously urban and cosmopolitan and determined to change the image of the party and purge that embarassing element from any view. The PQ logo itself was changed, marking this move, and Boisclair, like Bouchard, was economically not populist but neoliberal.

    Enter the ADQ which in particularly used the Herouxville controversy to speak of "reasonable accomodation", that immigrants should not impose their ways on the Quebec society that itself had forgotten its own traditions. The ADQ became the party of the nativists and xenophobes, cleaving them decisively out of the PQ orbit.

    When Boisclair left the PQ and Pauline Marois took over, she worked immediately to woo back this constituency with reasonable success. That combined with the poor performance of Dumont sealed the ADQ's fate.

  47. Point taken Eric. I will still disagree with you :) but you are probably on the right track not to mess up your model at this late stage. The plain truth is that we are seeing a sudden surge which, if it sustains itself to election day, would mean no current model is going to be really accurate other than the one relying on the last poll snapshots.

  48. Hi Eric, love your work. I'm not a Green supporter, but thought you'd appreciate this poll showing Elizabeth May set to win her riding:

  49. This very much has a 1990 feel to me. What is interesting is how the GTA talk radio is going nuts over it. 70 cent dollar, massive deficits, skyrocketing taxes, dogs and cats living together, total chaos!

    I'm wondering if this big surge could result in a squeezing out effect on Liberals, Greens, and BQ. NDP voters will feel like they have hope and a reason to go out and vote. Conservative voters will feel they have to stop the demon at the gate. Liberals and BQ though are just depressed and feeling like 'whats the point' while Greens are losing the protest vote to the NDP and having to count on those who actually prefer the Green platform (as I do) only.

    The Conservatives (& Liberals) have been pushing for a 2 party system via their policies and actions for awhile. They just might get it, but not in the way they hoped. The old 'be careful what you wish for, you just might get it'

  50. Éric,

    I'm a spreadsheet-oriented guy, and one thing that drives me nuts is that the poll tables under "Canadian Projection Details" do not appear to be sorted in a logical manner - not on pollster, not on date, not on weight... You would make this Excel guy VERY happy if you would sort the poll results either strictly on date or first on pollster and then on date.

    Pretty please? :o)

  51. Eric which set of EKOS numbers are you using? They released two sets yesterday, the top line numbers and the "absolutely certain to vote" numbers. Interestingly, Frank Graves backed the second set of numbers in yesterday's commentary, as more likely.

  52. A good ground organization is essential to identify your supporters and pull them on E-day. How do you do that when you've got little organization in many ridings but have a surge going your way? First, it appears people are motivated on their own to vote...Bloc, Liberal, and Conservatives pulling their supporters on E-day may, in fact, be getting NDP voters to the polling stations instead; the NDP is by far almost everyone's second choice .. it has the highest ceiling for growth. Also, consider the use of robo calls to all voters, social media, etc. Campaigns aren't what they used to be ... and consider what kind of a ground organization Rob Ford had in Toronto to get 50% of the vote for Mayor.

  53. Forum Research - 3,150 IVR interviews all done last night:

    Tories 34%
    NDP 31%
    Liberals 22%

    The ceiling for the NDP seems limitless - Forum projects 108 seats!!


    Whoever refuses to see what is happening is either dogmatic, a hard core partisan of some party or just uninformed. No disrespect intended to anyone

  55. Today's EKOS poll

  56. I think the model Eric is using is a good model. By ignoring short term trends and bumps that even out or dissipate, whether they be false or just temporary, you can get a better grasp of a result well in advance.

    That is, unless there's a late but genuine, (Even if only temporary) shift of large proportions. (Which may be happening) There's not much to be done about that if true because it is the Achilles heel of the design. I remember Nate Silver changing his model at one point in 2008... but for the American campaign he was able to take care of that well in advance of polls closing. Our campaigns are too short for such mid-campaign alterations.

    Eric; Maybe after the election we can see what your projection model would have predicted if it were based solely on more recent polls and see if that would have proven more accurate given these potentially strange circumstances.

  57. Eric,

    I don't know how far back your polling data goes, but have the NDP ever polled this high in the course of an election campaign?

  58. I'm not only enjoying your analysis this election, I've been using it as a tool to educate my kids (2 boys 15 and 20 and an 18 year-old daughter voting for the first time)about how our system works and what polling information does and does not tell you.

    I'm a bit surprised by how interested they have become in tracking the daily numbers and talking about it.

    I think one of the important sidebars to the polling is how difficult it is to use statistical models to account for instability. As you know, but for the sake of th general discussion, these sorts of models assume that their is a consistent underlying cause and effect relationship driving trends. When new trends with new causes jump up suddenly they don't do so well. When the trends are strong and contrary to what has gone before, older data biases the prediction andyour gut is as good a predictor as the models.

    On the other hand, if nothing else pops up to counteract the trends in play at the moment, you should have a pretty decent handle on what's happening with the numbers coming out between now and Sunday.

    Ground campaigns matter, and I am sure that there is a problem for the NDP in Quebec given the suddeness of their emergence, but it is important not to overstate these things. Whereas it was difficult, even 10 years ago, to get out the vote in a riding with few resources, modern technology has made it somewhat easier.

    The Obama campaign in the US showed very clearly that central campaigns can co-ordinate calling of identified supporters and get them to the polls with a small, professional local effort. I don't know to what extent the NDP is tooled up for this, but I would be surprised if any of the three federal parties isn't using the lessons learned in the Obama campaign to some extent. My guess is that the NDP will send in paid organisers to get the e-day efforts set up and run them over the phone and net from Montreal and/or regional centres. It's not quite as good as the real deal on the ground, but it is a lot better than the nothing many commentators are suggesting will be in place.

  59. @Anonymous (12:45)

    Nanos is not middle of the pack right now - not for Ontario at least. They are reporting the highest Conservative Ontario numbers out of any of the pollsters, and by a noticeable (but within MoE) amount over the next highest.

    @Anonymous (12:47)

    There's no saying the CPC internals are any more accurate than the publicly released polls are. Their numbers could be polling high (as the public polls did in 2004 and 2006), or they could be (foolishly) making several assumptions about the results that may prove false.

    Or it could simply be the old social engineering trick - act like you belong there, and people are more likely to believe that you belong there.

  60. @Anonymous 12:45

    I know Nanos did good jobs in 2004 and 2006 (as SES research) but you're still picking what you like and ignoring what you don't. The truth is that even in 2006 where it had very accurate results FEDERALLY (not always on a province-to-province basis), if you look at the daily polling, you will see huge differences from day to day.

    For example, in 2006, less than two weeks prior to election day, it predicted that Ontario would vote for Conservatives with 44% with the Liberals far behind at 35%. A few days before the poll, the Conservatives had fallen down in its poll to 39% versus 38% for the Liberals. Final results: 40% Liberal, 35% Conservative.

    So you can't just go by one of their daily poll to base your own prediction. As 2006 proves, they have had major fluctuations in the vote during their tracking poll, you have to try to discern between what is a real trend and what is just noise due to uncertainties. If you had picked the peak of Conservative lead in 2006, you would have expected almost a majority government thanks to 9 point lead for the Conservatives, which was miles away from the 5-point lead the Liberals finally ended up having.

  61. BTW, I forgot. The Conservatives ARE taking aim at the NDP. Harper's strategy had been to ignore the NDP, hoping they would rise a little, eating into the Liberals' support. But now he is making direct attacks against the NDP, he has even had TV ads made specially against the NDP. This is unprecedented.

    I think that they really have the internal polls and are scared that the NDP surge is eating into their possible majority. There is no other way to explain this complete shift in strategy.

  62. And Frank Graves hedges his own poll again today. He shows the Conservatives at 34% but oh ho, he ads this caveat.

    "Graves said the Conservatives have a two-point “hidden advantage” across the country because their voters are more enthusiastic, committed and certain to vote. This makes accurate seat projections tricky, he said."

    So he's backed off yesterdays seat projections and is really polling the Conservatives at 36% among sure to vote voters. You see he's coming within the margin of the other pollsters.

    He will show 38% among sure to vote voters before Monday, so he can claim he was withing the margin of error if they get anywhere from 35-41%.

  63. Kevin,

    On Sunday, along with the final projection, I'll run the numbers for only the last weekend of polls. Should give another perspective on a possible result.

    Anonymous 13:52,

    Not for a very long time. I think you have to back to the 1988 campaign.

  64. Regarding lack of NDP ground organization in Quebec, two points. First, 1200-1600 people attended Jack's recent Montreal rally. Second, and more fundamentally, I'm not sure the NDP needs to worry about pulling its vote in Quebec as the other three parties will be doing this for them. None of the other parties have the time required to sift through their identified vote to filter out NDP switchers.

  65. Yeah the 1988 campaign, when their polling numbers didn't materialize on voting day.

    Don't get me wrong, I believe the NDP vote will probably increase, but they're not taking 100 seats, not even close.

    Those numbers show the Bloc at 3 or 4 seats. This just isn't happening and absolutely NO riding polls have shown this.

  66. As far as comparing the Obama campaign to the NDP in Quebec um LMAO.

    The Obama campaign spent 1 BILLION dollars getting their voters to the polls. That's 1,000,000,000. Um the NDP in Quebec, not so much, maybe 1,000,000. That's three orders of magnitude less than Obama.

    And the unions in Quebec back the Bloc, and that doesn't change in two weeks of a campaign. Also people don't vote for Jack, they vote for their local candidate, some of whom don't even live in Quebec, most of whom are AWOL. And as far as any comparisons to the ADQ, yeah the ADQ campaigned out of garages. That's a whole lot better than being unavailable in Vegas and not living in the province.

    That will cause a dent in the NDP ballot.

  67. Not an expert in Canadian politics by any means, but I'm not sure the NDP - Lib Dem comparison holds. The Lib Dems are a much more ideologically vague party that lack a "tribal" vote. Isn't the NDP much more clearly defined as a center-left party with union affiliation? In this sense, they more resemble the British Labour party and the Liberal are more like the Lib Dems?

    Does the Canadian Liberal Party have a "tribal" vote?

  68. Jordan

    Of course the other parties have the time and organization to know who their supporters are. They are on the phone right now finding that out,and have been for the last two and a half years. And the Conservatives have the easiest time of it.

    The Big Bad Conservative Machine needs to focus on 10-13 ridings in Quebec. The Liberals many of the ridings, and most around Montreal. The Bloc, all ridings except some on Montreal's west island.

    The NDP is in the greatest position of disadvantage because they have to phone everyone in the province to figure out where they have a chance. They have no idea, no time and no resources to identify their supporters.

    The others know where their supporters are, the Cons with the easiest time of it.

  69. Anonymous 15:16,

    As I said in my post, I wasn't comparing the New Democrats and the Liberal Democrats politically, but rather in the similar role they play in our multi-party systems. They are both smaller third parties that haven't (at least recently, in the case of the LibDems) formed government before.

  70. Eric, I just wanted to let you know how much I've appreciated your blog over the past few months. I haven't commented before, but have been reading it daily for quite a while, and checking it several times a day since the election campaign began. Thanks for all your hard work.

  71. Anonymous said (27 April, 2011 15:04) ...
    "As far as comparing the Obama campaign to the NDP in Quebec um LMAO." etc.

    Apparently my original response to this derisory mischaracterization of my post was too tart for the tender sensibilities of the 308 community and failed to pass moderation. Je m'excuse.

    If you found my original post that amusing, you will no doubt be ROTFL over this one.

    To be clear, I did not compare the NDP campaign in Quebec with the Obama campaign. I noted that the Obama campaign brought many changes to election day organising that have implications for ground organizing for all Canadian parties and that will mitigate the problems with weak local organizational presence somewhat. This is a well studied reality and is covered in many places. There is a general treatment of the subject on Wikipedia.

    On a more nit-picking level, Obama did not spend a billion on GOTV. He spent $513,557,218 on his entire campaign, most of which was eaten up on the usual expensive advertising and organizational overhead. His GOTV exercise was innovative and tech-savvy and it is this that Canadian parties, including the NDP will presumably be trying to emulate.

    The social-activist community and union movement has split in this election, and some extremely well-tooled activists are working for the NDP. There is talk in the Quebec blogosphere of the Québec Solidaire organization backing NDP candidates.

    Bottom line - they will not be as entireley without an election day organization to get out their vote as might seem likely at first blush.

  72. Anonymous, I hear what you are saying, and you are right that a good organization on the ground is vital for winning most elections. However, in September 1990, people at the voting station where I lived were calling to their loved ones, "which one is the NDP candidate?" The union organization didn't bring those people to the polls.

    It isn't too hard to go out and vote, and if people are motivated enough, they'll go, and they won't need a union worker or some other campaign guy to encourage them to do so. Maybe in most cases, a lot of people will stay home, but if enough people just up and decide "I'm voting NDP today, the organization on the ground or the lack thereof isn't going to make one whit of difference.

    But I guess we'll see come May 2nd.

  73. Always an interesting read. My question is that with all these polling you show hardly any seat in Ontario changing hands -- are they projections only adjusted on a weekly basis? I understand that the data input would require strenuous amounts of time/manpower, so just curious as to why it appears every incumbent and party will be reelected in the province.

  74. Hahahahahaha.

    They did win approximately over a hundred.

    But I don't blame you, the polls were horribly inaccurate.


COMMENT MODERATION POLICY - Please be respectful when commenting. If choosing to remain anonymous, please sign your comment with some sort of pseudonym to avoid confusion. Please do not use any derogatory terms for fellow commenters, parties, or politicians. Inflammatory and overly partisan comments will not be posted. PLEASE KEEP DISCUSSION ON TOPIC.

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