Saturday, April 16, 2011

Week 3 Ceilings

The Conservative ceiling in this past third week of the campaign has risen to a very comfortable level for Stephen Harper. The Liberal ceiling, on the other, has dropped by 16 seats. 

The ceilings are established by taking the best regional results for each party from all of the polls released during the week, and running seat projections with those results. Of course, these calculations are greatly influenced by the smaller samples of regional polls. But we can still draw some useful information from these ceilings, as it is unlikely that the parties are capable of outpacing the best polls when you consider that the best polls are likely a few points higher than reality thanks to the MOE.

The Conservative ceiling is based on receiving roughly 46% of the vote in Canada. This number combines the best regional polling results for the Tories this week: 49% in British Columbia, 70% in Alberta, 59% in the Prairies, 49% in Ontario, 25% in Quebec, and 49% in Atlantic Canada.

With these levels of support, the Conservatives would likely win 25 seats in British Columbia, 28 in Alberta, 25 in the Prairies, 68 in Ontario, 12 in Quebec, and 19 in Atlantic Canada for a total of 178. That's well into majority territory, and in fact this scenario puts the Liberals only two seats ahead of the Bloc Québécois, 54 to 52.

This scenario isn't, however, extraordinarily plausible. While the results out West and in Quebec are not very unusual, a 17-point lead in Ontario is, and it is difficult to see the Conservatives reaching almost 50% in Atlantic Canada. But these levels of support do not put the Conservatives ahead in a lot of long-shot ridings, instead they limit the opposition to their fortresses. That is not such an unlikely scenario.
The Liberal ceiling is down from 117 last week, and based on a national haul of about 34% of the vote: 34% in British Columbia, 19% in Alberta, 23% in the Prairies, 42% in Ontario, 27% in Quebec, and 43% in Atlantic Canada.

This would give the Liberals 11 seats in British Columbia, none in Alberta, three in the Prairies, 45 in Ontario, 21 in Quebec, and 20 in Atlantic Canada for a total of 101. But the Conservatives would still win 135 seats, and the combined totals of the Liberals and NDP would still be four short of the Tories - five if we include André Arthur.

This scenario also counts on a few stretches: the NDP vote needs to collapse in British Columbia and Atlantic Canada, while the Bloc needs to drop to a historic low in Quebec. The results in the three Prairie provinces, Atlantic Canada, and Ontario are not unthinkable, and in fact don't even vary all that much from what some of the pollsters have been reporting on a regular basis.
If we look at how the weekly ceilings have changed since March, we can see that the Conservatives have been increasing their range - though not drastically. The ceiling has varied from 160 to 178 seats since March, a relatively minor variation that can be chalked up to the statistical wobbling of the polls. What is notable is that their upper ceiling has remained over 155 seats, which is not where I had their ceiling for much of 2010.

The Liberals, on the other hand, have varied a bit more, from a low of 86 in the first week of the campaign to a high of 117 seats last week. None of these ceilings have put the Liberals ahead of the Conservatives, which is where they need to be. They all, however, show that there is some room for relatively significant growth for the Liberals.

If you haven't seen them already, be sure to check Le Devoir for today's updated seat and vote projections. These do not include the poll released by Nanos this morning, though tomorrow's vote and seat projections that will be featured on The Globe and Mail's website will.


  1. Is there a direct correlation between week 3 higher CPC ceiling and the COMPAS (CPC biased) poll?

    The Liberal ceiling dropping with Nanos and EKOS polls this week has a greater significance to me.

  2. Hi Eric:
    Why no popular vote and seat projections update at the top of the page today? I wait to see this each day and it is still labeled as Friday.
    Thanx for all your work on this page. Great job.

  3. "This scenario isn't, however, extraordinarily plausible. While the results out West and in Quebec are not very unusual, a 17-point lead in Ontario is, and it is difficult to see the Conservatives reaching almost 50% in Atlantic Canada. But these levels of support do not put the Conservatives ahead in a lot of long-shot ridings, instead they limit the opposition to their fortresses. That is not such an unlikely scenario."

    Um...what?! Bottom line this. Is this plausible or not? I'm trying to parse "isn't, however, extraordinarily plausible" and I still can't figure out what you mean. I think you added one conditional too many. Otherwise, brilliant work!

  4. Comment from Frank Graves regarding his latest polling results;

    "At the end of Week 3, our tracking reveals clear patterns in the 41st federal election campaign.

    Despite the wildly inconsistent results that have come out of other polling organizations, we are very comfortable with our numbers and the well-behaved patterns that have emerged."

    Another Anon

  5. I posted mine on the blog, so not bad in terms of predicitons. I gave the Tories a floor of 113 seats and ceiling of 173 seats while for the Liberals one of 53 seats and 108 seats for the floor. Either way it seems about as likely as the Leafs winning the stanley cup next year that the Liberals will beat the Tories in terms of seats at this point. An arrangement or holding them to a minority and then winning the next election seems like there only alternatives if they want to govern before 2015. Nonetheless, I still think a Tory minority is more likely than a majority but a lot could depend on voter turnout. A low turnout should favour a Conservative majority while a strong turnout should favour a Conservative minority

  6. Hi Eric;

    Can you please post ceilings for the New Democratic Party, as well as for the Liberals and the Conservatives?

    Also, it would be interesting to know the combined ceiling for various combinations of parties. For example, does the ceiling for Liberals and New Democrats combined exceed 154 seats? I would imagine that many New Democrats (and many Liberals as well) would prefer a scenario where the Liberal-NDP total was a majority, to one in which their own seats were maximized but the total was less than that of the Conservatives.

  7. I didn't think the Conservatives were going to to get their majority. They may still shoot themselves in the foot. However the signs of desperation coming from the Liberals have led me to be somewhat more optimistic.

    I expect we will see a Conservative response to the $11 Billion cut from health care attack ad.

    As Iggy becomes more desperate he loses legitimacy in the eyes of Canadians. The more he fails the more he loses.

  8. Two signs point to a Tory Majority:

    1) the pro-Tory Trend in the polls: from this we can infer that a majority of those still undecided will incline more toward the Blue Team than toward any of the other parties;

    2) the absence of Tactical Voting. There is no noise about Liberals encouraging their sympathisers to vote NDP in ridings where only the NDP has a decent shot at beating the Tories; and vice versa.
    So in a lot of ridings, the anti-Tory vote will split, opening the way for very weaks Tory victories in a lot of competitive ridings.

  9. Anonymous 17:38,

    Projection updates are only on weekdays. On the weekends, you can see the latest updates in Le Devoir (Saturdays) and on The Globe and Mail website (Sundays). I'll be back with a new update here on Monday.


    I'm just running these ceilings for the Liberals and Conservatives because they have the most at stake, as well as because of time constraints. And finding the best combination of NDP and Liberal seats would require too many calculations, as I'd have to run every poll through the model (which, because of the way it works, isn't as automated as you might think!).

  10. The Liberals pulling out Martin and Chretien to campaign is a sign of desperation.

    Angus Reid was in the field to poll the debate. Would the Star decide not release the AR horse race numbers?

    An AR poll of 42 -22 -22 would push the Liberals over the edge.

  11. An excellent blog! - Congratulations and well done!

    A question, and perhaps this simply reflects my roots in polling coming from the United States, but I will ask it none the less...

    All of the polls you are using in your projections rely upon samples of all Canadians. But in the 2008 election, only about 59% of Canadians actually voted.

    In the US we just about always use "likely voter" polls where the pollsters screen the sample to try to reflect the subset likely to actually vote.

    For example in the most recent EKOS poll (currently the least favorable to the Conservatives with a smaller 7.5% lead) notes in their "Voter Intensity Index" that Conservatives are substantially more motivated, enthusiastic, and likely to actually vote than Liberals.

    Common sense would seem to dictate that this intensity factor be somehow factored in.

    Looking back at the the 2008 polls, compared to the actual Conservative margin of about 12%, I also note that every single poll I could found understated the lead the Tories had over the Grits...

    EKOS predicted a 9% gap (off by 3%)
    Nanos predicted a 7% gap (off by 5%)
    Harris/Decima predicted a 9% gap (off by 3%)
    Angus Reid Predicted a 10% gap (off by 2%)
    Strategic Council predicted a 5% gap (off by 7%)

    Taken as a whole, it seems the pollsters systemically under polled the Tories by 3% or so.... which seems a whole lot like the typical 3% gap between the Adult and "likely voter" polls in the United States.

    Your comment?

  12. We don't really have this issue fleshed out in Canada, we've always used decided and leaning voters. You're right that the Conservatives were under-estimated in 2008, but they were over-estimated in 2006. I don't have the numbers to check whether that was due to voter enthusiasm, as our polls are relatively simple and the data isn't very accessible.

    Because the pollsters were off in 2006 in a different way than they were in 2008, it is difficult to guess at how they'll be off in 2011. This voter enthusiasm could be the culprit, but with only EKOS tracking it (and Ipsos tracking a "absolutely certain to vote" group of voters), we don't have a great deal of data with which to work.

    The projection model does take into account some of this error, so hopefully it will turn out to be on the money.

  13. Hold your water folks. There'll be no Tory majority. The recent uptick is only that an unrelated uptick. The overall Tory trend will be downward. As I predicted at the start of the campaign the CPC will win a weakened minority then what they have now. The libs will boost their total by 10-20 seats, and with the polls as they are now who the heck knows what's going on with the NDP and Bloc.
    As for strategic voting, I expect you won't hear anything. It'll happen on election day, especially if there's any hope of a CPC majority.
    Spin away folks but the Tories are on the way out. They are the party of the past. Canadians are beginning their swing left and are leaving the CPC behind.

  14. Yes -- when is next Angus Reid due ?

    It's been a while.

  15. Éric

    Does your model consider Tactical Voting in some ridings?
    Where did you find data on the matter?

    And what about undecided voters? The last Nanos did evaluate them @ 15%. But in the cross-tabs of the other pollsters (excluding AR) I didn't find any info on the undecided-yet-likely-to-go-vote voters.

    Could you enligthen us on the matter? Was the percentage of undecided voters higher than 15% prior to the campaign? And which parties are they deciding to vote for?

  16. Katechon,

    "Does your model consider Tactical Voting in some ridings?"


    "And what about undecided voters?"

    No, they aren't taken into account. Generally speaking, undecideds break in similar proportions to decided voters.

    "Could you enligthen us on the matter?"

    Some pollsters report all of their findings, others don't. I'm not sure why, so I can't really speak to it.

    Part of it depends on the methodology. Online polls have very, very few undecideds, while IVR systems have fewer undecideds than traditional phone polls. And then prompted polls have fewer undecideds than unprompted polls.

    Undecided rates were usually between 20% and 30% before the campaign, so they have been getting smaller. Nanos had the undecideds at about 24% at the beginning of the campaign, but now they have them at 15%.

  17. Pinkobme said 'Spin away folks but the Tories are on the way out. They are the party of the past. Canadians are beginning their swing left and are leaving the CPC behind.'

    All the evidence points to the exact opposite to me. Iggy has totally failed to excite the electorate andhis 'Rise Up' rant just made him look like an idiot. Layton is eating Iggys lunch and the more of that lunch he eats the better things are for Harper. Lets face facts Iggy is not going to get anywhere near as many seats as Harper. In fact he'd be very very lucky to get 90 seats.

  18. Thanks Éric

    A poll was released today on the Undecided:

    "Among the 57 per cent of Canadians who said they were "absolutely certain" they were going to vote — not far from the 59 per cent who actually did vote in 2008 — the vast majority (84 per cent) indicated they've already made their choice.

    Six in 10 dedicated voters did so before the campaign even began, while 19 per cent had their minds made up before the debate, according to survey's findings. Just four per cent of Canadians locked in their vote after the debate, the poll results suggest.

    Of all eligible voters, not just those who are certain they will vote, the survey found a mere nine per cent remain on the fence, but Ipsos Reid CEO Darrell Bricker said politicians really need to be concerned about those who realistically make it out to the polls."

    "The survey found men, as well as older (55+) and middle aged (35-54) voters are more "locked-in" than women and young (18-34) voters and that NDP, Green and Liberal voters are more open to switching stripes or voting strategically than Conservative and Bloc Quebecois voters."

  19. There's lots of talk in the blogosphere about the so-called "vote mob" movement on Canadian campuses. If this spreads, and inspires more young people to get involved, how likely is it that the polls are going to capture the influx of newly interested voters. Are some pollsters better placed than others to pick up on unanticipated developments like this? Pollsters using a preconstituted participant pool would be particularly disadvantaged, I would think.

  20. Hey P: As I said above, the CPC will win -- a weakened minority. 90 seats puts the Libs in the zone I predict. Layton is a tremendous campaigner and I have great respect for him. But he's still about 10 points behind the Libs. Actually I hope the CPC hangs on to its 39ish stance till election day. It'll help the Libs greatly as Green and fringe NDP backers will vote Grit to stop a CPC majority.
    My comment about the CPC losing was intended to mean over time. They probably won't lose this election but will the next -- in one or two years.

  21. I second the request to get ceiling and floors for the NDP, as well as the Greens and the Bloc. I recognize that the math would be difficult to figure out, but I'm sure you've got an easy way to do it. I think it's time we give all of the parties a fair shot. Only giving floors and ceilings to the Conservatives and Liberals gives readers the impression that you are against the other three parties and it makes me question your credibility.
    I really liked Greg Morrow's democraticspace but I think his position as the campaign chair of the Green Party put up huge questions about his opinions and his credibility.

  22. It is interesting how far the EKOS seat projection varies from the one here. Is it reasonable to conclude that the higher ceiling scores this week are due to the skewed COMPAS poll?

    What would the numbers be without the Sun Media COMPAS poll?

  23. "An AR poll of 42 -22 -22 would push the Liberals over the edge"

    Hmm...sounds a lot like the AR poll WHICH ACTUALLY HAPPENED during the 2008 campaign:40-21-21. That one pushed the Liberals over the edge, right?

  24. "I predicted at the start of the campaign the CPC will win a weakened minority then what they have now. The libs will boost their total by 10-20 seats, and with the polls as they are now who the heck knows what's going on with the NDP and Bloc"

    Is there any basis (other than wishful thinking) for that prediction? Because, as Eric numbers demonstrate (you might have noticed them, he helpfully puts them at the top of the page),the "polls as they are now" as pointing to a stronger Tory minority (at the edge of a majority) and the Liberals losing seats (or maybe holding what they have).

    And, I gotta say, with the Liberals breaking out Paul Martin and Jean Chretien and launching their latest round of attack ads (inspired, no doubt, by the "men, with guns, in our cities" series of ads) the Liberals are steaming. They've peaked about two weeks too soon.

  25. Interesting news today that may contradict Eric's article in the Globe and Mail.

    We have the Liberals at 28.3% in the Nanos poll. That's actually their lowest result in all 17 days of Nanos daily tracking.

  26. Openisbetter,

    I understand, but my ceilings seek to answer two questions: can the Conservatives win a majority, and can the Liberals form government? While I could add the NDP, Bloc, and Greens to my calculations, I don't think it is as interesting an exercise. While I am sure some others disagree, they are free to make their own determinations if they wish. I only have so many hours in a day!

    As to your point about Greg, he has been critical of the Greens since he left so I think you should take him at his word. Like mine, his site is numbers-based. I can only believe that Greg is as dedicated as I am to let the numbers speak for themselves.


    I have the Liberals at 28.1% today in the G&M.

  27. Greg Morrow was hired by the GPC. He has, and always had his own interests. Besides, partisanship is different in the Green Party. People are only there because of their policy interests, so amongst Greens issues are what counts. Also, since Greens all come from some other political space, they are far more comfortable working with people who have different stripes. Like Greg when he came on board. I am absolutely certain Greg Morrow talks for Greg Morrow, and doesn't shill for anybody.

  28. Eric if the NDP do get really close to the Liberals in a couple of the polls will you do one of these for them ?

    Paul Wells had an interesting point the other day on twitter, which was that 3 days before the '06 election nobody could figure out where the 4th CPC seat in Quebec was gonna come from.

    Of course, we know they got a lot more than 4.

    Same idea with the NDP. Nobody can name where the 4th Quebec seat is supposed to come from.

    But we shouldn't be surprised if they get 8 seats !!

  29. I'd really like to see floors and ceilings for the NDP at least.

  30. Pinkobme - I would argue Canada is moving to the right much like the rest of the developed world. This has less to do with how people vote in the upcoming election but more looking at long term trends. The Conservatives are lot more right wing than the PCs under Robert Stanfield and Joe Clark were and likewise the Liberals today are not nearly as left wing as they were under Trudeau nor are the NDP as ideological as they were in the 60s and 70s. In the US the Republicans used to be fairly centrist unlike today where they are hard right and the Democrats used to be a truly liberal party not a centrist one as they are today. In the UK, the Labour Party used to be an unabashed socialist party and the Conservatives from Churchill through Edward Heath were pretty centrist. In continental Europe, the Liberals were social liberals not classical liberals like they are today and the Christian Democrats were centre-left who opposed the excesses of socialism not centre-right like they are today where they oppose the excesses of Anglo-Saxon conservatism. While the Social Democrats were truly socialists, not third way socialists who are happy to adopt centre-right policies when it makes sense. My point is regardless of how one votes, in terms of attitudes, Canada is moving to the right as are most developed countries. I would argue much of this has to do with lower voter turnout amongst the youth and the fact we have an ageing population as older people have always been more conservative than younger people. As for the Tories losing the election after this, I actually think a Tory majority is more likely to result in the Liberals winning next election than another Tory minority since they won't be restrained and are more likely to do something that would alienate some of their moderate supporters as well as it would give the Liberals time to rebuild instead of focusing on the next election. Never mind after 9 years in office, the time for change view will be more prominent than after only 6-7 years.

  31. Actually I hope the CPC hangs on to its 39ish stance till election day. It'll help the Libs greatly as Green and fringe NDP backers will vote Grit to stop a CPC majority.

    a 39 for the CPC from EKOS, HD and Nanos will translate to 42% on May 2.

    42 on May 2 makes it a CPC majority no matter how the collation vote migrates towards the Liberals (or NDP or Bloc)

  32. How come your model is so different from election almanac

    bq COn GRN LIB NDP
    2011.04.16 46 138 0 87 37

  33. Poll shocker!! Angus Reid out with a new poll in the Toronto Star and La Presse - NDP now TIED with the Liberals at 25% and tha Tories down to 36%(no chance for a majority at that level). Apparently Leger will also have a poll out tonight that contains some "surprises" - wonder if its the same surprise.!

  34. Yeah Carl I could be taking a big bite of wishful thinking these days. I still think this recent dip for the Libs is just that. You never know though. Iggy is still a rookie at this and has made some rookie mistakes, but overall has done quite well. It could be the hill is just too steep.
    Which brings me to Miles Lunn. Holy crap good analysis. I was speaking relatively, in the Canadian reality. As such social issues as health care become higher priorities in Canadians minds, they will shift relatively left in their standpoint.
    About three weeks ago I had a rant on here about how it might be best for all opposition parties, but especially for the Libs, that the CPC win a majority. I won't reiterate your points here, but I heartily agree. Nine years is about all Canadians seem able to stomach Conservative governments.
    If you take the long view of the shift to the right, I think an arguement can be made that there will be a necessary reaction which will pull it back left. If not, heaven help us.

  35. New Angus Reid

    Con : 36 (-2)
    Lib : 25 (-2)
    NDP : 25 (+4)


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.