Thursday, April 21, 2011

NDP gains in Quebec, but no new seats - yet

The Twitterverse was abuzz late last night, as rumours began being spread that CROP and EKOS would be releasing two polls showing the New Democrats ahead of the Bloc Québécois in Quebec. The gossip didn't disappoint, as CROP put the NDP at 36% and EKOS at 31% in the province, with the Bloc standing at 31% and 24% respectively. Talk of an NDP sweep of Quebec and of Jack Layton as Canada's next leader of the Official Opposition abounded. Even the staunchest critics of polls came out to marvel at what CROP and EKOS had to say, seemingly forgetting their prior warnings and cautionary notes.

Of course, these two polls are remarkable. And whether they reflect the eventual May 2 result or not, they will have a huge influence on the campaign in Quebec and in the rest of the country. There is nothing particularly problematic with either of these polls and they do speak to a completely identifiable trend in Quebec - the NDP is gaining and the Bloc is losing. So there is no reason to dismiss them, but there is some reason to hold off on the congratulatory toasts. I'll address these two polls, and the others added to this morning's projection, later this afternoon.
Undoubtedly, this projection update will be anti-climactic for some. But that should not be a surprise. The other polls added to the projection this morning (from Nanos and Innovative) did not effect any great change in the rest of the country, and Innovative's poll in particular (pegging the NDP at 16% in Quebec) put a damper on New Democratic gains, not to mention this week's Harris-Decima poll (15%).

But even without the Innovative poll, I would not have had the NDP at 30% in Quebec. The projection model has always been skeptical when it comes to new trends. Until this level of support becomes a consensus opinion, and not one that lasts for a few days, you can expect this kind of approach to big swings like the ones we've seen in the CROP and EKOS polls today. And I should emphasise that it is not me being skeptical, but the model itself. It is hard-wired to react this way, and the ephemeral Liberal lead in Ontario is a good example of why it is the proper way to go.

Nevertheless, there have been some changes at the national level. The Conservatives and Liberals are each down 0.1 point to 38.6% and 27.9%, respectively, while the NDP is up 0.3 points to 18.2%. That is a big jump in national support, and actually puts them even with their 2008 performance. I suspect they will surpass it in tomorrow's update.

Regionally there have been very few changes.

Except in Quebec, obviously. There the NDP has gained 0.8 points and now stands at 19.3%. Long-time readers of the site will know that a 0.8-point gain in a day is huge. And only a few weeks ago people would have laughed at the NDP being pegged even this high.

While the NDP gains, the other parties fall. The Bloc Québécois has dropped again and is now at 35% support, while the Liberals are down 0.2 points to 20.4% and the Conservatives are down 0.1 point to 20.1%. With a few more polls confirming the NDP as the top federalist option, the projection model will begin to reflect that as well.

However, despite this gain for the NDP there have been no seat changes in Quebec, or anywhere else. Getting over the bar of two seats in Quebec will require more gains for the party, but once they get there we should start seeing a few seats dropping here and there with every point gained.


  1. For good or bad, NDP support always tanks on election day. I'm old enough to remember talk of the NDP replacing the Liberals as official opposition during the 1988 election. I don't remember the poll numbers at the time but whatever was in the air with the polls evaporated by the time people cast their vote.

  2. Can you give us a preview?

    With the large changes in polls would it be informative to provide a version of the projection that's a little more present-biased?

  3. So long as your projection model uses a ratio method, your projections will be sound for the NDP in Quebec, even during a threefold vote increase.

  4. @ freshacconci
    It's a bit of a stretch to point to 1988 election results as a precedent for a general election in 2011. There are very few similarities between the NDP then and now, and indeed, the number of parties, modes and methods of campaigning, and the fact that people learn and adapt. We shall see whther the NDP support is solid in another week and a bit.

  5. Hi Éric,

    Just a quick question for you on the Quebec numbers. Given that the increased NDP support seems to be coming from the Bloc, does this imply that their support is dispersed throughout the province, as is the Blocs?

    While I read the polling firm releases, it's very difficult to get a good feel on the regional nuances.

    Could you perhaps also comment, on whether your methodology would reflect this type of behaviour? (ie. The NDP has significantly increased broad based support, that isn't focused in a small enough number of ridings to win them seats, but perhaps knocks off the Bloc in some of the Urban areas in favour of another Federalist party?)

  6. @freshacconci: Indeed. But that was the election where the NDP decided that polarization on the free trade issue would guarantee a PC win, allowing the Liberals to come back from behind by talking up the issue.

    In this case, the NDP surely can't match the BQ on the ground and would be expected to drop a few points as a result, but the BQ's only angle of attack is to accuse the NDP of being a centralizing "Canadian" party -- which will probably help the NDP in non-franco parts of Quebec, and also in the rest of Canada.

    Meanwhile, the Conservatives must be very happy with the latest polls, as a moderate NDP bandwaggon effect could cost the Liberals lots of seats in Ontario.

    Ah, if only we had a more rational electoral system, so people could discuss policy instead of vote-splitting and tactical voting...

  7. in 1988, there was no Bloc Quebecois, the right-wing option was the *Progressive* Conservatives, and the Liberals were the party of Trudeau.

    things change. there's people who are saying "NDP is the official opposition" and "NDP Quebec Sweep?!" things don't change THAT fast. but things change.

  8. Why are you giving today's CROP poll (n=1000) less weight than a Harris-Decima poll (n=490) from April 3? The chart below shows CROP with a 0.98 weight and the H-D poll at 1.01.

    Eric, with respect, you are being cautious to the point of irrelevancy. I'm sure by June you'll show the NDP picking up 10 seats in Quebec!

  9. Éric,

    You may want to include these riding polls done in NL for NTV/Telelink in your next update if you have not already:

    (There is video there from the TV newscast)

    Avalon in particular is quite different from what your model is presently suggesting.

    Love the site, BTW, keep up the good work.

  10. Ok, I want to see the Bloc support collapse as much as anyone else, but 24%? That's really stretching it.

    Though, I do think the main three parties are handling the Bloc fairly well: ignoring them. (aside from the separatist jabs) That, and the new attack line from the Bloc on Jack is priceless: "He's a nice guy, but vote for me". Way to go handing over the votes.

    I'm not sure if anyone has polled for this, but the hardline stance on separatism coming out of the PQ convention followed by Gilles making stronger statements about it is likely pushing away any nationalist/soft federalists who voted Bloc in the past. Before, they were impotent on the separatist file, so a "thumb your nose" at the other parties style vote seemed safe. Now, he's looking a little more scary while everyone else is looking normal.

    At what point do we start wondering about strategic voting? Would we be able to see anything in the polls just before the election to postulate that it's going to happen? There doesn't seem to be an effort in place yet, but I can see a large number of Bloc ridings that could fall if enough voters decided to shift their votes.

  11. Anonymous 10:02,

    The NDP's support does seem to be across the board in Quebec. CROP has them at 40% in Montreal, 34% among francophones, and 28% in Quebec City. That probably puts them ahead in a few of the Montreal ridings that are close.

    The model does not take these sub-regional polls into account, however.

    Anonymous 10:14,

    CROP has not had a great track record of late. People should keep that in mind - this is the second time in this campaign that CROP has released something at odds with other pollsters (36% for the NDP this time, 11% for the Liberals in mid-March).

    However, your question made me double-check my numbers. The poll you were talking about from HD was weighted incorrectly, as I had the sample size twice as much as it was in reality. That has reduced it to about 0.70, though it doesn't have a big effect on the projection.

  12. Hi Eric,

    What I love about your site is the caution - very...Canadian.

    I am curious however how you anticipate the model will work in the final days before the election? Right now your model, as I understand it, seeks to confirm trends over time and over pollsters. The temporal component hits a wall on May 2nd when the vote comes and, presumably, your model doesn't require post-election day polling to predict vote % or seat counts for e-E-day.

    Do the last few days of polling somehow get "super weighted" in your model?


  13. Highlander,

    I've done some tests using the polls from 2006 and 2008 as my models, and the system I am using for this election was the one that performed best.

  14. Éric,

    What actual data was used to drive up the NDP support to above 23% in the Pontiac riding? This would indeed be historic support.


  15. Could you give us a projection for the seat distribution in Quebec based on this CROP poll? Just to see what would happen if this was actually the result in Que on E-Day

  16. James,

    I threw one out on Twitter last night: 31 NDP, 30 BQ, 10 CPC, 4 LPC.

  17. The NDP surge in Quebec seems to have taken off in the last two nights. Any polls older than that won't reflect the change.

    What's happening is very interesting if it has reached critical mass. Pundits think it is being fueled by a backlash against Marois saying they were going after another referendum and FRENCH Montrealers deciding to vote NDP, as well as the Anglos on the plateau (Mulcair)

    That being said riding polls two days ago showed the NDP up only 5% in the Quebec city region, which only helps the Conservatives keep their seats.

    We'll see. I'm still not convinced. These polls certainly get them some seats in the Outaouais and Montreal. I'm not sure about anywhere else.

    What is more interesting is if it rubs off on the rest of the country, where many people have been holding their nose and voting Liberal because they didn't think the NDP was viable. It also excites their voters enough to come out.

    If it's sustained, its a game changer. I don't think it affects the Conservatives much though.

    I don't usually do this kind of crystal ball gazing, but this shift has the potential to foretell a collapse of the Liberal vote and reduce the Liberal Party to a rump.

    We'll see.

    btw, Forum has a poll out showing the NDP in second place nationally.

  18. In politics, things stay the same until suddenly they change. We love them until suddenly we hate them. The PCs found that out in 1993, the Libs in 2006 and now the Bloc might be getting a taste of that this year. Maybe not. But I wouldn't cry over it if the Bloc lost 10 seats or so -- especially to the NDP. With a slightly stronger performance from the Libs this time out (compared to the last election which was a fiasco)I expected the NDP to get shoved out to some extent and maybe end up with a seat tally around 25. That doesn't seem to be happening, even though the Libs are doing relatively better. Hopefully this is at the expense of the Bloc. I respect Duceppe but I'm not fond of separatists -- in any part of the country. They're like two-year-olds throwing a fit in a busy store hoping to embarrass mommy and daddy into giving them what they want. Only two-year-olds have an excuse for their infantile behaviour. And I'm including western types in this as well.
    Anyway, as usual we'll see what this all means on May 2. This little twist just makes things all the more interesting.
    Thanks Eric for your great work, as usual.
    I'm outta hear for the Easter weekend, and (blasphemy of all blasphemies) won't be following the polls much until the middle of next week.
    Happy Easter everyone. Have a good one.

  19. "The survey of 1,000 Quebecers online took place between April 13 and 20. Due to its non-random character, there is no margin of error."

    I must say, I am really dubious of this at the bottom of the CROP poll.

  20. On that Forum poll, and its Quebec results:
    "the poll found NDP support has mushroomed to 34 per cent, with the Bloc Québécois second at 25 per cent, the Liberals at 18 per cent, the Conservatives registering as a preference for 16 per cent of voters and the Green Party with four per cent."


  21. The FORUM poll also has the Tories up 42 / 28 over the Grits in Ontario, with the dippers at 20%

    42/28/20 gives the Tories better than 60 seats in Ontario....

  22. I'd be interested to see a riding poll on Mount Royal. I think it is probably not a Liberal Romp as Eric shows it. If 20 points have shifted from Liberal to the NDP its a three way race there.

    The effect of an NDP surge may in fact be to win or secure more seats for the Tories.

  23. Another strange and interesting tidbit from the CROP poll.

    "The company surveyed 1,000 English and French speakers in Quebec, and the data were weighted for gender, age, language, and "social/cultural values."

    Weighting for Social and Cultural values????? What?? That's what the polls are measuring.

    So did they ask other questions first and them decide how to weight the vote question? Ouch! The poll is garbage if that's the case.

    Ekos also asks other questions first. The Forum Poll was done on one night only, Wednesday, too hot.

    So each of these staggering polls has some questionable methodology, asking questions first, online pre-selection for "cultural values" weighting, and only polling one night.

    I'd take them with a grain of salt. I don't doubt that the NDP is surging in Quebec, just how much is the question, and I don't think it's this much, for now.

  24. @Anonymous (11:21)

    It's kinda true, but very poorly worded.

    There is no accepted method of calculating margin of error for non-random surveys. The line could be fixed by simply adding "available" to the end of it.

    For what it's worth, the original La Presse comment on margin of error can translate as "This online survey does not include a margin of error given the non-random character of the sample". Somewhat more accurate wording than the G&M.

  25. Just looking at BC, an NDP surge loses the Conservatives one seat but gains them two others in Eric's model.

    In the model, Cons lose Burnaby Douglas but gain Esquimault Juan de Fuca and North Van. Net benefit Tories plus one since they already hold North Van.

    The NDP can pick up Kingsway and Burnaby New West from the model. In the other the numbers just aren't there. As they already hold these two seats it's a wash.

  26. In Alberta, Saskatchewan, and Manitoba nothing changes in the model with an NDP surge. In the North Nunavut could swing NDP if there were a huge surge.

    So to the border of Ontario, we have a net gain of Conservatives plus 1 for an NDP surge and one toss up.

  27. To the East, because I'm saving the big ones for last, an NDP surge nationally produces for:

    New Brunswick: +2 Conservative
    Moncton River Dieppe

    Nova Scotia: 1 toss up from Conservative
    South Shore St. Margarets

    PEI: 1 toss up from Liberal

    Newfoundland-labrador: +1 Conservatives, 1 toss up from Liberal
    Random-Burin_St. George, Con
    St. John's S. Mount Pearl, toss up from Lib.

    So for the East the grand total is +3 Conservative with three ridings moving to three way races, 2 from Liberal, 1 from Conservative.

    The Grand total for the Country excluding Ontario and Quebec with an NDP surge

    +4 Conservative, +4 ridings move to toss up three way races.

    The NDP actually make no gains outside of Quebec and Ontario with a surge of about 8% in overall voting intentions, which is quite a jump.

  28. Sorry in BC the NDP hold Burnaby Douglas in a surge changing in the model, not Burnaby New West.

  29. There is no way that the NDP will take Nunavut. The election there is not so much about the party but the candidate. Unless the cabinet minister (Cons) and the former premier (Libs) split the vote, it is unlikely the NDP will take this seat.

    And I am going to vote for the NDP there.

  30. Anonymous said . . .

    "Weighting for Social and Cultural values????? What?? That's what the polls are measuring."

    Actually these polls are primarily measuring vote intention. If they can find measures of social and cultural values that are reasonably stable over time (which vote intention certainly is not), that's a reasonable thing to weight for. I'm guessing that this Quebec poll is using some sort of orientation towards separatism (hard separatist, hard federalist, in-between category(ies)) that could be pretty stable.

  31. Now for Ontario, wait for it folks,and NDP surge of 10% gains the NDP a total of 0 seats on Ontario in Eric's model.

    It does however gain the Conservatives 7 seats:
    Kitchener Waterloo
    Don Valley West
    Eglington -Lawrence
    York Centre
    Mississagua South

    It also moves 5 ridings into toss up category (my own designation not Eric's as they are too close for me to tell what Eric's model would say) from Liberal to Conservative:
    Etobicoke Centre
    Richmond Hill

    So not including Quebec, the grand total for an NDP surge of 8-10% nationally is Conservatives +11, toss ups + 9 (only Eric doing these calculations can say how these seats go. I can't do them with a rudimentary shift of votes to the NDP from the Liberals.) The NDP gains 0 seats on the win by a mile model but may gain some of the toss ups.

    This my friends gives the Conservatives their majority, if they hold 8 of their 10 seats in Quebec. 147 +11 is 158, plus they would win a few of those 9 toss ups in Eric's model.

  32. I just have a couple of questions. The first is, if the NDP are at 40% support in Montreal as has been suggested, would it be Liberal or Bloc seats in Montreal that could be in play?

    Also, the article on the Forum poll has been talking about an NDP official opposition. Don't you think that NDP support has been too broad while Liberal support is more centered in areas like Montreal and Toronto and wouldn't that hinder NDP support from actually converting into seats? I'm thinking of the recent election in the UK where Lib Dem support started to surge only to have a lackluster finish.

  33. According to Ekos, NDP up 12.5% since Monday in ... Manitoba and Saskatchewan . Where did their added support come from? Cons down 12.5! (Word of caution, margin of error for Sask/Man in the poll is a whopping 8.9%)

    Doug Johnson Hatlem

  34. Oh crap, I just lost my Quebec post with all the riding changes, and the grand totals. I don't know if it got sent, and it was difficult. Take Quebec numbers with a grain of salt, they really are too complicated to estimate.

    The grand total nationally for an NDP surge of 10% was Conservatives +12, NDP +4, (Libs +1 but they lost so many this doesn't count), with 28 seats in close races that I can't calculate.

  35. The NDP as opposition has more to do with a Liberal collapse in seats than an NDP gain. I think they only gain 12 seats and become opposition because the Liberals hemorrage.

  36. Irish Observer21 April, 2011 13:42

    Surely if (and it's obviously a massive if!) the NDP took 40% in Montreal they'd win a large number of seats on the Ile? You'd have the other 60% doled out between Bloc, Liberals and Cons - except in highly Liberal or highly seperatist ridings its unlikely either would get over 40%, thus giving the NDP wins by coming through the middle?

    40% tends to to be the sweet spot in multi-party First past the post systems at which a majority of seats fall (see Conservatives at a national level). If the NDP really did manage such dense and strong support they could be in for a major seat wave in Quebec.

  37. To the Anonymous who's done the "what if the NDP got a real surge?" analysis over a few posts here: Doesn't it massively depend on where that NDP support comes from? There are valid arguments for its source being Conservative (I have a union card in my wallet and a gun rack in my truck), Liberal (I have a Blue Cross card in my wallet and a Prius in the driveway), Green (I don't have a wallet or a car), and not-previous-voter (Holy crap! You mean my vote for the NDP might actually mean something? I haven't voted since university). All of those mean different things, and are likely spread out over different ridings. We have data about Quebec - it seems that your comments are correct there, because they're drawing from the Bloc. However, they might draw directly from the Conservatives in some areas, or even motivate new voters.

    How did you do it for your projections?

  38. That's ridiculous. So the NDP needs pretty much 75% of the national vote to gain a seat? Sure, that sounds credible.

    About as credible as saying the Conservatives can win 155 seats with less than 40% of the vote.

  39. Dan weighting for whether or not someone is a separatist is absurd and grossly skews this poll as to make it garbage.

    You can't weight for something that is changeable as in an opinion.

    Polls can only be weighted for age, gender, region, native status etc. and other things that don't change.

    If crop is doing this kind of weighting then the numbers should be given separately, such as saying "separatists support the bloc by....."

    They can decide how many separatists or federalist there should be in a sample and then weight it. It makes the numbers a poll of a poll.

  40. Frank, I can guarantee the Conservatives can win 155 seats with less than 40% of the vote, especially if the Bloq, NDP and Liberals split the vote much like the Reform Party, Bloq and Progressive Conservatives did in 1997.

    Liberals 155 seats, vote 38.46%

  41. @ Frank
    With vote splitting on the left the Tories can win 175 seats with less than 40%.

  42. Ashley Morton, of course it depends where the NDp support it, however what I was trying to point out is that most ridings just won't be affected by and NDP surge.

    In those that will be affected, more move to the Cons than the NDP. This is true no matter how the vote is distributed.

    The biggest vote distribution question for the NDP is in Quebec where 17 ridings move into the possible category. Will they win most of these? Nope. Why because the NDP surge is concentrated in Montreal and that is where they have the organization on the ground.

    I just did an across the board vote switch to see what would happen. It's very crude but clear in most ridings.

    I think the best the NDP could do is 12-15 more seats nationally, (about 10 from Quebec) which would put them as official opposition. However, the Tories would have their majority by taking seats from the Liberals because their vote went to the NDP.

  43. Actually, I allowed a 20% shift in votes in Quebec, which is massive. Bloq down 10%, NDP up 10%. The numbers just aren't there for an NDP surge to get them many seats. They just hurt the Liberals and the Bloq mostly.

    And Eric's model is much more resistant to change than this simple shift.

  44. Irish observer, not necessarily. Montreal is divided into English and French enclaves. They are also polling the greater Montreal area, not just the island. The French vote in Montreal is moving to the NDP, not the Anglo vote. The Liberals will hold most of their seats. The NDP may take some Bloq seats, or some races may become competitive or even go Conservative or Liberal in aka Brossard, Westmount, Ahuntsic.

    We'll see.

  45. New Democrats are taking votes from everyone. Models that try to put voters on a left-right spectrum spent too much time in poly sci class and not enough time actually talking to voters.

    The two main parties are both full of scandal. People are holding their nose and voting Conservative just as much as people are holding their nose and voting Conservative. It's the corruption and broken promises that have driven votes for the past 30 years, ignoring the conservative split in the 1990s.

    Quebec has "held their nose" more than anyone. Yes, you have the obvious "anyone but Harper" voters who pick from the Bloc or the Liberals. But then you have people who hate the Liberal corruption, so they vote Bloc, even though they're not separatists. And then there are people who are STRONG federalists but hate the Liberal party, so they vote conservative.

    In Quebec, it's not about left-right. It's just as much about corruption and federalism, maybe even more so.

    Also, the "left-right" analysis breaks down when you look at individual ridings.

    The places where the NDP run the strongest are actually Conservative-NDP ridings (70 to 100 of those), not Liberal-NDP ridings (only 30 to 40 of those). Academics think it's about policy. But it's really about who you are and where you're from. These are either rural communities based on mining or farming, or they're towns where the economy is centered on a big auto plant or other factory, with a strong union. Voters in these areas are attracted the NDP on economic issues, but are also attracted to Conservatives on social issues (and just overall being pissed off at the Liberal scandals).

    So I wouldn't be making predictions of an NDP surge at the expense of Ignatieff. The Liberal-Conservative races are mainly in the Ontario suburbs, where there's no NDP surge to speak of.

  46. There is no "Q" in "Bloc".

  47. anonymous,

    What I see is people saying the NDP are up by x% so I will that x% to each NDP vote total from the previous election.

    That method is fine as an intellectual exercise but its meaningless. Everybody's vote tends to pool here and there. There is simply no reason to assume that the NDP vote will not act like everyone else's.

    Instead we hear that a 15% vote for the Conservatives will produce 5 times more seats than 35% will for the NDP. Or that a 3% fall for the Conservatives and a 8% rise for the NDP will mean the Conservatives gain seats.

    Sure, if you simply add x% to 2008 vote totals I'm sure it might look that way. But at some point common sense has to enter the fray.

  48. Ah no q in Bloc. Thanks Eric.

    I don't think the NDP is taking any Conservative votes to speak of, given you could serve tea off the flat line 39% the Conservatives are polling across the board, across time, across polling firms. It's rock solid. Scary rock solid.

    That being said, as a Liberal, I was going to vote Conservative and I am now considering voting NDP, but it has to do with my riding and the fact that I want a purge of the Liberal Party old boys.

    ps. If Coderre loses his seat to the NDP, it would be sweet revenge for what he did in Outrement. That's an inside joke for non Iggy Liberals

  49. Have any polls been conducted in Esquimalt Juan de Fuca (BC)? Your projection shows a Liberal win there, but the Liberals only did so well there in past elections because of the popularity of Dr. Keith Martin. With him not running, I expect this year it will go to the NDP candidate, Randall Garrison, who ran in 2006 and beat the Tory Candidate (who is now on his third attempt) by a margin of 31.3% to 27.5% (Martin won with 34.9% of the vote, just 3.6% ahead of Garrison). Similar results in 2004: Garrison came in 2nd with 30.6% (to Martin's 35.3% and 24.2% for the Cons). Granted, in the last election the NDP's candidate, Jennifer Burgis, only got 22.7% of the vote (Cons had 34.1%, Lib 34.2%), but with the popularity of the NDP generally, Garrison back as the NDP candidate, and a lesser-known Liberal candidate...

  50. Frank Graves was on power and politics and said that a new poll reconfirms today's result, that they didn't even believe.

    Also, he said all the support is coming form the Bloc.

    You can listen to the podcast on the CBC web site.

  51. "The French vote in Montreal is moving to the NDP, not the Anglo vote." (21 April, 2011 14:34)

    Correction: according to that poll, it seems that the NDP are getting more of the non-franco votes than the franco ones (percentage-wise: 45% vs 34%). Almost half of the non-francos are anglos anyways, not just allophones.

  52. Andrew Drummond21 April, 2011 17:19

    I think your projections are missing the effect of Hec Cloutier in Renfrew-Nippissing (pardon the spelling).

    There are people talking as though he could win, and even if he doesn't he certainly eats into Gallant's numbers a bit.


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.