Wednesday, October 28, 2009

New Ekos and AR Polls

Two new polls for your consumption: EKOS and Angus-Reid.

First, EKOS.

Taken between October 21 and October 27 and involving 3,220 Canadians, here are the national results:

Conservatives - 38.4%
Liberals - 26.8%
New Democrats - 16.7%
Greens - 9.9%
Bloc Quebecois - 8.2%

Nothing much new in these numbers. In British Columbia, the Conservatives are at 36.8%, followed by the NDP at 28.9% (very good) and the Liberals at 25.0% (good). The Greens are at 9.3%.

In Alberta, the Conservatives lead with 62.9% followed by the NDP at 13.2% and the Liberals at 13.1%.

In the Prairies, it's 51.9% for the Tories, 24.6% for the Liberals, and 18.5% for the NDP. A weak number for the NDP here.

In Ontario, the Conservatives have 41.4% of the vote, the Liberals have 31.2%, and the NDP has 15.5%. Again, nothing much new here.

The Bloc seems to have taken a hit in Quebec, and stands at 33.5%. The Liberals follow with 23.8% and the Conservatives with 22.5%. The NDP is at 11.6%.

In Atlantic Canada, the Liberals have 38.0%, the Tories have 32.3%, and the NDP has 22.1%.

Of note, however, is that the Liberals have the lead in Vancouver (33.6% to 32.2% CPC), Toronto (37.6% to 36.5% CPC), and Montreal (29.2% to 28.6% BQ).

The poll would result in the following seat totals:

Conservatives - 147
Liberals - 83
Bloc Quebecois - 46
New Democrats - 32

The poll also asked whether each party leader should remain as leader or be replaced. Stephen Harper got 45% support to stay, while 40% said he should be replaced. The split was 31% to 46% for Michael Ignatieff and 51% to 25% for Jack Layton. Obviously, Layton has the best number here, while Ignatieff's is troublesome.

More troubling, however, is when this question is broken down by party support. 85% of Conservatives think Harper should stay, while only 10% think he should be replaced. 72% of New Democrats think Layton should stay, while only 17% think he should be replaced. But only 55% of Liberals think Ignatieff should stay compared to 26% who think he should be replaced.

Now, Angus-Reid.

Taken between October 23 and October 24 and involving 1,001 Canadians, this poll found the following national support:

Conservatives - 40%
Liberals - 26%
New Democrats - 17%
Bloc Quebecois - 9%
Greens - 7%

Very similar to the EKOS findings. The regionals, however, are quite different.

In British Columbia, the Conservatives have a decent lead with 43% followed by the Liberals at 27% and the NDP at 25%. The Greens had only 2%.

In Alberta, the Conservatives are down to 52% while the Liberals are up to 23%. The NDP is at 9%.

In the Prairies, the Tories have an unbelievable 73%. The Liberals have 11% and the NDP has 9%.

In Ontario, it is 41% CPC, 31% LPC, and 17% NDP. Similar to EKOS.

The Bloc are doing well in Quebec in this poll, with 40%. The Conservatives are next with 21% and the Liberals are close behind with 20%. The NDP is at 15%.

In Atlantic Canada, the Conservatives lead with 35%, the Liberals are next with 32%, and the NDP is in third with 26%.

This poll would result in the following seat totals:

Conservatives - 152
Liberals - 78
Bloc Quebecois - 52
New Democrats - 26

Phew! Now that this is all out of the way, I'll do a projection update tomorrow morning.


  1. Those AR Prairies numbers are a thing of beauty.

    There are just certain people in politics who annoy me and Ralph Goodale is one of them.

    I think there's a real shot of beating him next election. Then I think it might be possible to mop up the last of the Liberals in Manitoba, though the NDP has some pretty entrenched members there.

  2. Am I the only one that finds it interesting that the poll includes the "Undecided" numbers for the questions about leadership, but not in the overall polling numbers?

    Makes me wonder why they are trying to skew the overall numbers. "Undecided" has been pretty high in past polls... if those numbers were included (the way they should be) the numbers would be quite different.

  3. Eric compared to LAST year's official election results for each political party is the variance today statistically --->significant or is it with the normal margin of error to make a statment?

    Groundhog day


  4. Anonymous,

    Polls about vote intention are usually reported based on decided voters only (either decided or decided plus leaning)

    In vote intention questions the parties are being measured simultaneously so there is no additional information about the relative strengths of the parties being lost. I.e. the voter is not more undecided about some parties than others.

    However, when questions about leadership are asked about each leader separately, then "I don't know" answers are also being accumulated separately.

    This can make a difference.

    If leader 1 shows results:
    positive: 15
    negative: 30
    undecided: 55

    while leader 2 shows results:
    positive: 30
    negative: 60
    undecided: 10

    In this case a failure to report undecided numbers hides some very crucial information, even though both show the same positive-negative split. I don't think that there is any attempt to "skew"; the pollsters are just following a logical and conventional methodology.

    With regard to your comment: ' "Undecided" has been pretty high in past polls '

    The undecided level of 16.4% on the vote-intention question reported by ekos is not particularly high -- indeed it is well within the normal range.

  5. Given Ekos's methodology (automated polling done with push-button answering to fixed options) would tend to reduce the undecided vote.

    I also think it would tend to inflate Green support. If you contrast Ekos numbers with Nanos numbers, for example (since Nanos does not prompt, forcing you to come up with the name of the party you support yourself), you see that Ekos routinely reports Green numbers roughly double those that Nanos reports.

  6. Of course, Éric has already built the various polls' biases into his projection model.

  7. Martin,
    I agree, but I would at least like to know what the undecided number is.

    The last Environics poll showed undecided at 29%.

    That's a pretty significant number and would translate to a much different outcome if an election were held.

  8. Anonymous wrote:

    "I agree, but I would at least like to know what the undecided number is."

    Fair enough. I agree that pollsters should report the undecided number -- even if only parenthetically or in footnote.

    "and would translate to a much different outcome if an election were held."

    How do you know that? Given two sets of identical results among decided voters but with different numbers of undecideds, we have no way of knowing that they would point to different outcomes (or, if different, in what way it would be different)

    Pollsters typically assume that, on most questions, people who are undecided are either:
    1) not going to vote anyway (remember that about 40% of eligible voters did not vote in the last election) OR
    2) of those who will, that they will wind up breaking out in approximately the same proportions as the decided voters

    While there is no compelling logic that dictates #2 is a valid assumption, historically, assumptions 1 & 2 together seem to hold up in terms of polls and actual elections.

    Also, although Environics' undecided number was a little on the high side, it is noteworthy that their party numbers are very much in line with the other pollsters like ekos.

  9. More of the federalist vote in Que. is turning Conservative then I would have thought.

    Vive la Canada!


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.