Monday, April 25, 2011

NDP bon deuxième in two Quebec riding polls

There's only one national poll to review today, but there are two riding polls that have been added to the mix over the weekend. They were both taken in Quebec and they both show that the New Democrats are competitive. However, aside from the polls for Gatineau and Outremont we have yet to see the New Democrats leading in any individual Quebec riding polls. Might they be the runners-up across the province?
First, the Nanos poll. This poll was taken on April 21, 23, and 24. If we compare it to the last complete Nanos poll, taken April 18-20, we can see that the Conservatives are still holding steady. They are up only 0.2 points to 39.2%.

The Liberals, however, have dropped 1.1 points in the past four days and now stand at 25.6%, only two points ahead of the NDP. They have gained 1.5 points over that time period.

These variations are not statistically significant, but they are part of a discernible trend.

In Ontario, the Liberal vote has collapsed. It has dropped 7.1 points since April 18-20 and now stands at only 29.3%. The Conservatives are up three to 47.8% while the NDP is up 2.3 points to 16.9%, still below their 2008 performance. With a split like this, the Conservatives will have no problem winning a majority.

In Quebec, the NDP's rise is palpable. They're up 6.8 points to 30.2%, followed closely by the Bloc at 27.4% (-4.6). The Liberals are up 1.2 points to 22%, while the Conservatives are down 3.4 points to only 14.1%. That is a very low mark for them, though with the Bloc as weak as it is the Tories should still be able to elect most of their incumbent MPs.

Looking at Nanos's daily tracking charts, we can see that the Liberals have been dropping steadily for almost a week in Ontario and Atlantic Canada. The Bloc has also dropped for five straight days. The NDP has been rising in Ontario, but they still have some ways to go before they can win new seats.

The same goes for the NDP in Quebec, at least according to two new riding polls.

The first was conducted by Segma Recherche for La Voix de l'Est, a newspaper based in Granby. It shows the Bloc's Christelle Bogota (who left the NDP shortly before the campaign began) leading with 32%. That is only a drop of three points since the 2008 election, however.

The New Democrat Pierre Jacob is not far behind with 26%, an increase of 17 points since the last election. Apparently the NDP's vote grew with each day of the poll and by the end of it Jacob was leading, but we're talking a daily sample of 100 or so people.

The Liberals, at 26%, are also in the race, but that is a drop of seven points. The Conservatives are also down seven points to 11%.

But this poll does show how the New Democrats are competitive in ridings that would not have been considered likely pick-ups for them before the campaign began. But the Bloc vote appears solid in the riding, despite the provincial drop in support.

The projection was within four points for the Bloc, Liberals, and Greens in this riding, but underestimated the NDP vote by 10 points. This could be a common refrain on election night, but the NDP still has some ways to go in my projection model. We just need a few more days.
The other riding poll was conducted for Chambly - Borduas, a riding east of the island of Montreal. In this poll, Yves Lessard of the Bloc has a comfortable lead with 37% support but that is a drop of 13 points since 2008. The New Democrats have taken up some of the slack, and their candidate Matthew Dubé is running second with 24%, up 10 points since the 2008 election.

What is most interesting about this riding is the presence of an independent: Jean-François Mercier. He's a well-known actor/comedian with a somewhat rough sense of humour. Nevertheless, 15% of the riding's residents intend to vote for him. According to the poll, about half of his vote comes from former Bloc supporters.

The Liberals are running third with Mercier at 15%, generally where they were in 2008. The Conservatives are down about eight points to 7%.

Aside from Mercier, the projection was good for this riding. It was only off by one point for the Liberals and the NDP (that is an especially good sign), though it was off by eight points for the Bloc and six for the Conservatives. That is mostly attributable to the unpredictable nature of Mercier's candidacy.

These polls tell two different stories for the Bloc but just one for the NDP. In Brome - Missisquoi the ability of the Bloc to hold on to its vote is demonstrated, while in Chambly - Borduas we see Bloc supporters jumping to a slightly ridiculous option. But combined with the Bloc's dropping support in Brome - Missisquoi over the four days of the sample, we can speculate that the Bloc's support is proving to be very soft. Perhaps after voting Bloc for almost two decades many Quebecers are starting to become curious about other options.

And in both ridings we see that the other option is the NDP. They are the most popular federalist option in these two ridings, a real strength considering that the NDP appears to be drawing support from social democratic sovereigntists (or soft nationalists) as well. And when we consider the margin of error, the NDP has a good shot in Brome - Missisquoi and several of the other ridings that have been highlighted in these riding polls.

On the other hand, the NDP has yet to be shown in the lead in a riding other than their two main targets: Gatineau and Outremont. Getting 20%-30% of the vote in a dozen or two ridings might be very likely for the NDP on May 2nd, but getting over that 30% mark and actually winning could be too much for a party with a weak ground game in the province.

The NDP's main opponent in Quebec has a relatively well-oiled machine, and has the apparatus of the Parti Québécois from which to draw. The Conservatives were able to win seats in 2006 out of the blue, but the provincial Liberals were helping them and it is far easier to elect government MPs.

The parallels between the Tory breakthrough in 2006 and the potential NDP breakthrough in 2011 might be more limited than it appears on first glance.

43 comments:

  1. Eric when is the latest a poll can be released? Is it Thursday at 11:59 pm or Friday at 11:59 pm?

    And would that be Atlantic time or NFLD time, so it could be 10:30 EST?

    ReplyDelete
  2. My understanding is that the rules have been changed to just 24 hours, so the last polls would be released on Sunday.

    However, I may be wrong and would appreciate being corrected if I am.

    ReplyDelete
  3. "With a split like this, the Conservatives will have no problem winning a majority."

    God help us.

    ReplyDelete
  4. Eric, your skepticism of the possibility for NDP gains in Quebec is not based on the numbers. The four most recent distinct (not double counting Nanos) polls in Quebec all show the NDP ahead of the Bloc, and three show them over 30%. Plus all show NDP up trends, Bloc down trends. If the NDP finishes first in the Quebec popular vote it will win lots of seats. For it not to finish first will require a shift back away from it, not mere stasis, certainly possible, but far from certain. Based on the dates and the reported trend, the NDP has a very good chance of winning in Brome-Missisquoi -- especially if you read the original report, the poll is not evidence of Bloc solidity. You may be right about the ground game, but the polls themselves are very encouraging for the NDP.

    ReplyDelete
  5. They most definitely are, Dan. I don't doubt that for a minute.

    ReplyDelete
  6. Eric,

    Fine analysis as always - much appreciated. As we speak, Innovative Research has a new poll out showing the following for Quebec:

    NDP 36%
    Bloc 27%
    Conservatives 18%
    Bloc 14%

    The Liberals especially seem to be all over the place in Quebec - very difficult to determine what their trending is like.

    ReplyDelete
  7. The Elections Canada website states the restrictions on opinion polling as: "No new opinion survey results may be published on election day before polling stations close."

    http://www.elections.ca/content.aspx?section=med&document=mar2911&dir=pre&lang=e
    (Dated Tuesday, March 29, 2011)

    ReplyDelete
  8. Environics 25 April

    Tories 39
    NDP 25
    Liberals 22
    Bloc 7
    Greenies 6

    ReplyDelete
  9. "Eric,

    Fine analysis as always - much appreciated. As we speak, Innovative Research has a new poll out showing the following for Quebec:

    NDP 36%
    Bloc 27%
    Conservatives 18%
    Bloc 14%

    The Liberals especially seem to be all over the place in Quebec - very difficult to determine what their trending is like."

    Could you post a link to that poll?

    ReplyDelete
  10. @Skoblin:
    Bloc at 14%? I doubt it is wrong.

    ReplyDelete
  11. The Bloc at 14% isn't nearly as problematic as the bloc at both 27% and 14% in the same poll.

    ReplyDelete
  12. I presume the Innovative Research poll mentioned above should read the Liberals at 14% instead of the Bloc. I haven't located a reference to it yet, but will comment once I do.

    ReplyDelete
  13. This comment has been removed by the author.

    ReplyDelete
  14. Chambly is the only other riding which has ever gone NDP in Quebec (won by Phil Edmunston in a by-election in the late 80s)

    ReplyDelete
  15. Eric, it's looking more and more like your final prognostication will be for a Tory majority (gulp). Anyway, regardless of whether it's minority or majority, will your data-crunching allow you to offer a probability of such a result? Obviously, the goal is to be 100% right, but I wonder if it won't be close enough that you'll be offering us a probability of such a projected result, or just a flat prediction. Thanks.

    ReplyDelete
  16. I won't be calculating any sort of probability, but my final projection will discuss more than just what the model spits out.

    ReplyDelete
  17. You should take these three Newfoundland and Labrador polls into account in your projections.

    For one, the riding of SJS-MP is more likely too close to call at this point. Your projection for Sullivan is way too generous.

    ReplyDelete
  18. They already have been taken into account and a post written about them.

    ReplyDelete
  19. Frank Graves is projecting 131 seats for the Conservatives and 100 seats for the NDP. Meanwhile every other pollster shows the Conservatives cruising to majority or extremely close to it. Dream on.

    ReplyDelete
  20. Hi Eric,

    My apologies if this is something you've been asked and answered previously in comments, but I'm curious how you're planning to assess your model.

    By my count, it makes a total of 1968 predictions. While that's a very good year, it's a heck of a lot of predictions! Indeed, we'd expect nearly 100 of them to be "significantly" wrong, just on general principles.

    Do you have some markers already established for yourself?

    Thanks,
    Chris

    ReplyDelete
  21. To the Anon who posted at 17:52, could you please provide a link to Frank Graves' projections? Thanks!

    Also, thanks to Eric for his great blog posts, I have been checking out 308.com since the election began.

    ReplyDelete
  22. Ekos's poll is from the weekend. The other polls are older...Ekos may be catching the wave and it will be confirmed by others.

    ReplyDelete
  23. Anonymous said:
    'Meanwhile every other pollster shows the Conservatives cruising to majority or extremely close to it. Dream on.'

    You might want to look at the polls more closely... Only the Ispos-Reid poll of last week showed the Conservatives in majority territory, the Nanos polls show are arguable at best for the possibility and all the other polls show diminishing prospects for a Conservative majority.

    Colin

    ReplyDelete
  24. Yeah... the new EKOS. I must admit I have considerable difficulty believing that poll. (especially since they're always showing the CPC 5+ less than everyone else) Big sample size though.

    ReplyDelete
  25. Chris,

    The main measure of success will be calling seats correctly. Bare minimum would be 90% accuracy. Next would be to have relatively correct seat totals. After that, I'll be assessing where I went right and where I went wrong in each riding and looking for general explanations of what worked and what didn't.

    ReplyDelete
  26. The link to the EKOS poll is:

    http://www.ekospolitics.com/wp-content/uploads/seat_projection_april_25_2011.pdf

    There is also an EKOS seat projection at:

    http://www.ekospolitics.com/wp-content/uploads/seat_projection_april_25_2011.pdf

    I have to say that the sample size appears beyond reproach and the implications very significant for all three other parties!

    Could we be looking for a scenario similar to when Rae won in Ontario?

    ReplyDelete
  27. I do think Éric deserves a bit of leeway on election night if his predictions end up further off than usual. This has been an odd week or two, and the result may yet be even odder.

    ReplyDelete
  28. If you go to http://en.wikipedia.org/wiki/Opinion_polling_in_the_Canadian_federal_election,_2011 and sort by Conservative vote, you'll see the three lowest results are all from Ekos. The lowest of all is today's Ekos.

    What do you guys think of that?

    ReplyDelete
  29. I really have trouble believing some of these NDP numbers, especially the Ekos poll, even if I'd like them to be true.

    That being said, there truly may be something there if we look at how political leaders are acting. All party leaders, who have their own polling done, much more frequently than what we have, have started to open fire at the NDP. Ignatieff and Duceppe, we can easily understand... but even Harper has, and he should be the last one to want to do that, considering the conventional wisdom that the NDP steals votes from the Liberals and splits the vote. He has basically ignored the NDP during the entire election and now he's blasting them... something has changed, something that has made the Tories realize that they needed to take aim at the NDP to get their majority.

    ReplyDelete
  30. Ipsos just put out a poll of 900 voters in the Lower Mainland of BC. It shows the Tories at 42% (Down 1%), NDP 29% (up 5 points) and Liberals 23% (down 1%) and Greens 6% (down 3). These numbers would almost certainly indicate NDP holds in Van Kingsway and Burnaby-Douglas and pickups in Surrey North and Newton North Delta.

    http://www.ipsos-na.com/download/pr.aspx?id=10676

    ReplyDelete
  31. Ontario Dipper: New Ekos poll out today places NDP at over 28% while the Conservatives have dropped to under 35%. Projection gives the Dippers 100 seats. In Quebec, NDP leads 38% to 24 % for the Bloc, while the Liberal and Conservative support has almost disappeared.

    Frank Graves refers to it as "terra incognito"

    Indeed!

    ReplyDelete
  32. Ekos's poll is from the weekend. The other polls are older...Ekos may be catching the wave and it will be confirmed by others...

    Ummmmmm.. NANOS polled 1/3 of his sample yesterday.. and it showed a mild tory uptick....

    ReplyDelete
  33. Good luck incorporating that massive and geographically wonky lower mainland BC poll from Ipsos!

    Doug Johnson Hatlem

    ReplyDelete
  34. But I thought on Debate night how amazing Layton was and I sensed a certain je ne sais quoi about him. He seemed vibrant and strong and sharp. I thought then maybe there was going to be "something" big about him. He's my pick for winner of the election. Even if he only gets Opposition, he's still the winner. Even if he only grabs a few more seats in Quebec, he's a winner. I've been amazed by him this campaign.

    ReplyDelete
  35. 53 NDP seats in Quebec according to Ekos makes no sense whatsoever. It needs to be adjusted for the concentration of vote in favor of the Liberals, Conservatives and Bloc in certain riddings (and for the NDP in at least 3 riddings). This would give a better distribution for Quebec:

    12 Conservative instead of 4
    9 Liberals instead of 3
    23 NDP instead of 53
    Still 0 green
    30 Bloc (thats a minumum, they might take more from the NDP)
    Still 1 independant (A. Arthur)
    Total 75 MPs

    Now, all things being equal in the rest of the country, the National distribution would be more like this:

    Conservatives 139
    Liberals 68
    NDP 70
    Bloc 30
    Other 1

    ReplyDelete
  36. I think something the pollsters (and this model) might be missing is the extent to which strategic voting is a factor for a very large number of people. The anti-Harper sentiment is so strong that even while I canvass for one candidate (who has little chance of winning) people are asking me who they should vote for to make sure to beat the Tory incumbent. Factor in the 8 different strategic voting websites, and I suspect that any projection model is going to be way off. We are going to see some *very* strange results on election night, that will seem to in no way reflect the conventional wisdom about various ridings and regions. While I can't say what the new government will look like, I can tell you its a whole new ballgame now.

    ReplyDelete
  37. 308 seems way behind with all new national polls showing the NDP at mid 20s and a new Ekos poll with a low margin of error and high sample showing the NDP at 28 per cent and now in a race with the Tories. Yet incredibly this site has the NDP at 19 percent. I may ignore it and something is definitely incorrect here.

    ReplyDelete
  38. Reminder re Toronto Mayoralty race this past fall: Mid-October Nanos Ford 43.9, Smitherman 40.5; EKOS Ford 48, Smitherman 33; actual result Ford 47.1, Smitherman 35.6

    If they get this one right too, everyone as in everyone will have to follow their methodological transformation.

    Doug Johnson Hatlem

    ReplyDelete
  39. Goaltender Interference26 April, 2011 09:33

    I unfortunately don't have first-hand sources, but it appears that a riding poll was conducted for the Tories in Brome-Missisquoi, showing the NDP in first place and the Liberals 20 points behind:

    http://www.lerefletdulac.com/%C3%89lections/Place%20au%20d%C3%A9bat/2011-04-25/article-2452751/Un-sondage-des-conservateurs-place-le-NPD-en-tete-dans-BromeMissisquoi/1

    If this is accurate, the NDP Quebec breakthrough could be less like the 2006 Tories and more like the 1984 Tories.

    ReplyDelete
  40. Anonymous 7:44 - you're clearly missing both the purpose and the methodology of Eric's site.

    It doesn't sway in the wind - trends have to be sustained before they're shown as the gospel truth here. That's part of why it's valuable.

    Don't forget that while 72 hours is an eternity to political watchers, it's not REALLY that long for a "flash in the pan" trend. It's getting there, but it's not there yet. I appreciate Eric's "conservatism" in not jumping in bed immediately with new worldviews.

    ReplyDelete
  41. Methodological transformation???? EKOS asks issue questions before the vote intention question. All they have to do, is not ask those questions first for the last poll before the vote to get a more accurate result. Then say oh look our methodology works.

    ReplyDelete
  42. One thing that strikes me about polls and about the work that you are doing, Éric, is that these statistical analyses are by their nature conservative and cannot really account for unpredicatable (but real) phenomena such as sudden shifts in voter sympathies. I'm not sure any model could account for such phenomena, but when it seems demonstrably to be happening, there must be a way of acknowledging it...

    ReplyDelete
  43. Ashley,

    Not missing the purpose or methodology (or importance) of the work Eric is doing here. It's why I follow it religiously, as I did with 538.com in the previous two American elections. The point of these things, however, is to do the damndest to get things right. Pointing out that EKOS was the only one to accurately predict a late and strong surge to Rob Ford in Tornto's mayoralty race is very important in such a case. If it turns out that EKOS gets things right in surge situations where old school polling models fall flat on their face (as would be the case if they all stay around where they are and EKOS gets things right), then that has to be weighted strongly in future models. 538.com got things basically exactly right in 2008 in the US pres. election and pretty darn good as well in 2010. That's what we want. But then, you have to discuss things like which pollster in the field is doing better and therefore deserves greater weight.

    Doug Johnson Hatlem (AKA Anonymous at 7:44)

    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.