Thursday, February 24, 2011

Tories drop in new EKOS poll, fall back to earth in Ontario

Once again, EKOS is out in front with a poll showing a new shift in voting intentions. When EKOS came out with a 12.5-point lead for the Conservatives two weeks ago, they were roundly criticized and their results questioned.

At the time, I argued that EKOS's results were not unreasonable, and a bevy of other pollsters came forward to confirm what EKOS was seeing. Now, as the polling firm pegs the lead at only 5.1 points, I can't help but wonder whether EKOS is going to be the first to identify a new shift again or whether the other pollsters will not confirm EKOS's findings this time in the coming weeks.Compared to that last poll, the Conservatives have dropped 4.9 points and now lead with only 32.4%. The Liberals, at 27.3%, have gained 2.5 points.

This IVR telephone poll has a margin of error of +/- 1.9%, 19 times out of 20. That puts these shifts outside normal statistical variation.

The other parties are wobbling within the margins, however. The New Democrats are up 0.6 points to 14.8%, the Greens are up 1.2 points to 11.9%, and the Bloc Québécois is up 0.6 points to 10.5%.

Note that, despite the large "Other" response (3%), that is no different than last time.

It can be posited that this drop could be the consequences of the Bev Oda affair, or it could be because the Conservatives tend to fall away from majority territory as soon as they come close to it. We can't really say, though Angus-Reid did find that 58% of Canadians believe Mrs. Oda should resign from cabinet. However, Angus-Reid neglected to first ask whether Canadians were aware of the controversy, spurring a few from the Ottawa press gallery to suggest that Canadians would demand the resignation of Len Blork, were they asked.

You can follow Len Blork, Minister of Grains and Oil Seeds (a plum portfolio if there ever was one!), on Twitter here. Despite the allegations, I'm projecting that the incumbency effect likely means he can hold on to his Nipawin seat.

Anyway, drilling down into the regionals we see that the main source of the Conservative drop has been in Ontario (MOE +/- 3.3%), where the party has handed over the lead back to the Liberals. They stand at 36.4%, up 6.1 points, while the Conservatives trail at 35.9%, down 5.6 points. The New Democrats are up 0.6 points to 14.2%. The Liberals are leading with 44.6% in Toronto, followed by the Tories at 31.9%. In Ottawa, the Conservatives are leading 39.6% to 38.5%.

In Quebec (MOE +/- 3.5%), the Bloc has gained 1.2 points and leads with 39.9%. The Liberals are up 2.6 points to 18.8%, while the Conservatives are down 3.9 points to 16.2%. It appears that this drop, outside of the MOE, is noteworthy. The New Democrats, meanwhile, are down 0.5 points to 11.4%. The Bloc leads in Montreal with 39.1%, followed by the Liberals at 22%.

The Conservatives are down 4.5 points to 32.9% in British Columbia (MOE +/- 7.1%), while the New Democrats are up 6.7 points to 25.2%. The Liberals have dropped 5.6 points to 20.9%, within striking distance of the Greens, who are up 5.9 points to 19.4%. All of these variations, however, are within the MOE. The Conservatives are leading in Vancouver with 39.3%, followed by the Liberals at 24.2%.

In Atlantic Canada (MOE +/- 7.5%), the Conservatives are leading with 36.8% (+4.7). The Liberals trail with 30.2% (-6.9), while the Greens are up 7.6 points to 16.8%. The New Democrats have dropped to fourth, down 4.6 points to 14.3%.

The Conservatives are down 12.3 points in Alberta (MOE +/- 7.5%) to 52.1%, trailed by the Liberals at 24.4% (+9.9). The Greens are up nine points to 17%, while the NDP is down 5.3 points to 5.7%. The Conservatives are leading in Calgary with 59%, followed by the Greens (!) at 20%.

Finally, in the Prairies (MOE +/- 8.3%) the Conservatives are down 2.5 points to 42.3%, but still hold a good lead over the NDP at 27.8% (+9.9) and the Liberals at 21.4% (-4.4).

With this poll, ThreeHundredEight would project 18 Conservative seats in British Columbia, 26 in Alberta, 20 in the Prairies, 46 in Ontario, six in Quebec, and 12 in Atlantic Canada for a total of 129. That is a drop of 22 seats from EKOS's last poll.

The Liberals would win six seats in British Columbia, two in Alberta, four in the Prairies, 49 in Ontario, 13 in Quebec, and 18 in Atlantic Canada for a total of 94, an increase of 11.

The Bloc Québécois would win 55 seats in Quebec, a gain of one over two weeks ago.

The New Democrats would win 11 seats in British Columbia, none in Alberta, four in the Prairies, 11 in Ontario, one in Quebec, and two in Atlantic Canada for a total of 29, a gain of nine seats.

The Greens would win one seat in British Columbia.

For regular readers, note that the new projection model is up and running for the British Columbian projections. All other regions are using the older model. I am open to individual, specific questions about these riding projections for British Columbia.

Will this poll dampen electoral speculation? Perhaps among the media, but at this point the party leaders probably have a good idea whether they are going to go into an election or not, and unless the polls become horribly bad (or terrifically good) for one of them I don't think they will be changing their minds. Political parties may not make their decisions based on public polls, but public perception is important, and opinion polls do play a role in shaping public perception of whether a particular political party is a "winner" or a "loser".

In the end, we will need to wait and see whether any other polling firms confirm this Conservative drop. There are some very high Green results in this poll, which probably need to be chalked up to the MOE. But since the overall shift in favour of the Greens is only to the tune of 1.2 points nationally, their results should not be considered a major factor in this shift.


  1. What are the seats numbers for BC in your old model?

  2. For this poll, I would have projected 18 Conservative seats, nine NDP seats, eight Liberal seats, and one Green seat in British Columbia with the old projection model.

    In other words, two fewer seats for the NDP and two more for the Liberals.

    That is not to say the old model is wrong - at the next election I'll calculate my projection using my old system to see whether it would have been better in the end.

    I am more confident in the new system, however, and it also allows me to make riding-level projections, which is a major plus.

  3. I think the overarching problem with Ekos is that by prompting for both Green and Other - they consistently grossly overestimate Green and Other support. As has been pointed out - in the last election when the Green Party was at an all-time never to be repeated high point - the Green/Other vote was 8%. This poll has it at 15%. Does anyone seriously think the Green party will get anywhere remotely close to 17% in Atlantic Canada where they are totally moribund and have averaged about 3% in recent provincial elections and byelections?

    If you compare Ekos and Nanos, FWIW, the Liberal number is almost identical - but nanos has Green and other at 5% instead of 15% and virtually all of that difference goes to the Tories and the NDP.

  4. The new model is riding-specific, right? So you can say which riding the Greens would win? Is it Gulf Islands-Saanich?

  5. The fact that you are able to report on this positive Liberal Poll on a Thursday is amazing.

    EKOS and CBC release the 10+ CPC lead poll at 6 pm on Friday.

    Not hard to see the bias there.

    You would think after being openly questioned about integrity and manipulation that EKOS would try to set some standards and stick to them.

  6. BCVoR,

    The poll two weeks ago was released on a Friday because of the Egyptian Revolution. CBC Power and Politics devoted their entire two hours to covering Egypt, pushing the EKOS poll to Friday.

    This one was bumped forward to Wednesday, according to EKOS, for scheduling reasons. That just might mean someone's birthday on Thursday that Frank Graves couldn't miss.

    In the future, conspiracy theories are best left to the dark corners of the internet, emphasized with CAPS and multiple exclamation marks.

  7. Eric, would you say the Elizabeth May win in this projection is mainly because of their national swing in support, which this poll has at higher levels than ever before and certainly outside of reality, or because of your riding level projections?

    Also, once your riding level projection model is complete nationally, will you release those full detailed results each time?

  8. Eric, you know the problem, don't you? Everyone whose not a Conservative is actually in the media. And and they control all the polling companies! Minus Allan Gregg who is a Conservative... but we all know he's a CINO anwyays!?! Agh!

    Anywho, look forward to seeing the new model into action for the rest of the provinces. I'm wondering whether you'll put up riding-specific results for us all to see, or is it a secret? (No doubt a secret part of the Vast Liberal Conspiracy!)

  9. Just out of curiosity, I'm guessing the Green seat in BC is because of a Elizabeth May effect, am I right? Just out of curiosity, do you mind telling us how much "boost" you give the Green party because of the presence of the leader? Because the Green party was not even that high in this riding last election. And you you assume that this "boost" or May-effect is at the expense of all parties or more from the Liberals and/or NDP?

  10. Bryan,

    The Greens win in Saanich - Gulf Islands because of their big number in British Columbia in this EKOS poll. I don't have the Greens winning in my actual projection, however, just when I input EKOS's numbers.

    No, I will not release full riding-level results for each individual poll. They will be released for ThreeHundredEight's "official" projection. That is why I say in this post that I'm willing to answer questions about the projection for individual polls for individual ridings. Posting full listings for each individual poll would be overkill. But, as I said, they will be released for the site's projection.


    There are a suspiciously high number of exclamation marks in your comment.

  11. When I next update the projection (scheduled for the Monday after next at this point), I will provide full details of the riding-level projections for whatever provinces have been added to the model by that point.

  12. Bryan B.,

    Yes, Elizabeth May is given a boost in the riding. The Greens were at pitiful levels of support in Central Nova in 2006, but she boosted their numbers in 2008, over-and-above what the lack of a Liberal candidate would have accounted for.

    She did the same in her first by-election attempt, when the Liberals did present a candidate.

    Her boost is based on the amount of vote she took from the Liberals and the NDP in that by-election and from the NDP in Central Nova (a similar amount). She also receives the "star candidate" boost because of the extra efforts the Green Party has promised in Saanich - Gulf Islands.

    Finally, the riding's projections are not based solely on 2008's result because the NDP's candidate dropped out.

  13. Oh I'm totally with you on boosting May, it's is obvious she can get more votes for her party than a regular candidate, as seen in 2008 or the previous by-election. I was just curious at how big this boost was. Because I still think May is making a mistake by running in this riding instead of one in Ontario for instance.


  14. sounds good. I look forward to seeing your riding-by-riding site projections!

  15. Eric:

    Bravo on where conspiracy theories belong!

    Besides, we all know the Big Media favor the Conservatives! Half of the Conservative senators used to work for CTV! And half of CTV want to become Senators! Thus their bias.

  16. Can you at least get the statistics right? The MOE for a change of a proportion is not the same as the MOE for one proportion in one poll. And that's not the same as the change in the difference of proportions (the lead) from one poll to the next.
    Have a read of:

  17. Volkov and Eric you guys seem to be rather naive today.

    What BCVOR is saying has extreme merit.

    Pollsters release these polls for free to the media for the national exposure.

    Of course Frank Graves is trying to drive the national conversation.

    The more attention (and even controversy) a pollster gets the better chance they have of getting paid contracts.

    (Whether this move had anything to do with his well known hostility to the Conservative party is debatable.)

  18. As always the projections reflect how fptp is nothing like true democracy (one person = one vote = equal influence). In BC with 19% the Greens get 1 seat, but the Liberals with 21% get 6, 25% gets the NDP 11 seats and 33% gets the Conservatives 18 seats.

    So to compare...
    19% = 1 seat
    +2% = extra 5 seats
    +4% = 5 more seats
    +8% = 7 more seats

    Yeah, yeah, it is the system we are stuck with so stop complaining right? But I bet a lot of current Conservatives complained a lot during the 1993-2004 period when false majorities happened every time. Of course, dummy me was a Reform/PC/Alliance voter (depending on election) back then and now I'm a Green supporter. Guess I'm good at finding the party most screwed over by FPTP.

  19. FC,

    There's no need to be rude, but thank you for that link. It would appear that everyone (myself, media reporters, even pollsters themselves) uses the MOE inaccurately, or at least slightly inaccurately. While not extremely accurate, it would appear that using the MOE in the way that I do (and everyone else does) is a good rule of thumb, as the differences highlighted in that paper are not terribly significant.

    I will try to be more careful in the future, however, and play around with those formulas.


    Obviously, pollsters are trying to get coverage. That Graves and EKOS plotted the Egyptian Revolution in order to move their poll release to a Friday evening two weeks ago because it was good for the Conservatives seems slightly unlikely.

    In fact, showing a huge Conservative lead would get a poll tons of coverage, so why someone would want to bury it is beyond me.

    And whether a Wednesday is better than a Thursday, however, is debatable as well. News tends to get lost in the middle of the week. Arguably, a Thursday is better as the poll is unlikely to get swamped by other news on a Friday or on the weekend.

    But, frankly, I'm simply tired of the conspiracy theories thrown around when it comes to pollsters. Believe it or not, few in the media or among those running a polling firm have a political bias that motivates how they approach their jobs. If that was such an important thing to them, they'd become politicians.

  20. "But, frankly, I'm simply tired of the conspiracy theories thrown around when it comes to pollsters."

    As am I Eric and thanks for that. Ekos is paid by the CBC on a continuing contract if I'm not mistaken so that negates that conspiracy theory.

  21. "Believe it or not, few in the media or among those running a polling firm have a political bias that motivates how they approach their jobs."

    That's a stretch. We've seen in the United States a history of polling being politicized.

    And in Canada we do have a number of firms that work with a specific party on and on going basis.

    So your point would be more credible if you distinguished here between neutral firms and those allied with specific parties.

    As for the media, I don't think there are many people who intentionally think "how can I frame this story to hurt/support party x,y,z".

    However, there are a great deal of people who have trouble being objective and who's political bias (because we all have one) seeps out.

    All in all I think its very healthy to be a skeptical and on the look out for bias in our polling and news media.

    We should be informed consumers.

    Not trusting and docile sheep.

  22. It could be the recent Ekos offering is somewhat out of touch, or it could be the first to show a change in voters' attitudes. I favor the latter.
    The CPC attack ads, I beleive, went a long way to helping boost the Conservatives' fortunes in the past two weeks. However, they're apparently a two-edged sword. It could be Canadians got tired of the same-old, same-old from our government and thus the Ekos results. The Bev Oda fiasco could have also played a role in the dramatic decline for the Tories.
    I'm curious to see if other polsters come up with similar results in the next week or so. I expect they might.

  23. There's being critical and informed, and then there is the witch hunt and the automatic assumption that someone who disagrees with you is obviously a hack from the other party.

  24. Shadow as to your last post about politicized media, I agree.
    However, in my experience, personal and observational, it tends to favor the Conservatives a lot more than any other party. That's likely because most media outlets are owned by huge corporations which like the tax breaks and other perks Conservative governments, including this one, is famous for.

  25. Pinkobme these are some pretty wild swings for a country that supposedly isn't paying attention to poltics, no ?

  26. Shadow,

    Apparently you and Frank Graves agree on something there.

    Liberal Conspiracy!

  27. By the way, I know the reason for EKOS's scheduling change this week and it is 100% coincidental and benign. Sometimes people just aren't available.

  28. Hey Shadow, you're right. For the CPC to go from 35ish to 39ish to 32ish in a couple or three weeks is pretty wild, especially when most of us are politically asleep. Perhaps though the constant CPC attack ads and the election speculation woke some of them up. Or maybe it's just the usual whobble in the wheel of Canadian political opinion.

  29. FC,

    I've turned the formulas in the paper you provided a link to into something I can use on a regular basis. I will make reference to this in the future, so thanks for bringing it to my attention.

  30. John Northey - I'm quite glad that our electoral system differs from true democracy. True democracy effectively means unlimited majority rule, and that would be terrible. That would allow the majority to do whatever it wanted at the expense of the minority.

    True democracy is incompatible with individual freedom.

  31. Volkov c'mon now, with Sun TV on the way you don't you'll ever be calling dirty tricks on anything ?

    I'm of the belief that there is no such thing as a bad question and that its always a good idea to be skeptical of what we see and hear.

    Blanket dismisal of people like BCVOR is uncalled for.

    An attitude of 'Nothing to see here folks, move along' is damaging to our democracy.

    Comments likes BCVOR's should be evaluated at face value (or ignored if you wish).

    Mockery is not an intellectually honest response.

    Eric pointing out that he knows the reason behind the scheduling change puts the matter to rest.

    The intial and uninformed response of attacking BCVOR, however, was the wrong approach.

  32. You might have a point, if this kind of "the pollsters are Liberal hacks" sort of talk wasn't so common. You'll forgive me for growing tired and frustrated with it.

    BCVoR's uninformed attack on EKOS was the wrong approach. His comment was no example of reasoned skepticism.

  33. I hear just as much "the pollsters are Tory hacks" about Angus Reid and Ipsos Reid.

    Rasmussen and PPP in the states are always the subject of back and forth.

    As someone who followed the 2010 US midterms I saw A LOT of junk polls.

    Judging someone's skepticism as "reasoned" or not in a false distinction.

    We'll need all the skepticism we can get to defend us from the darker side of the polling industry and keep our democracy strong.

    BCVOR pointed out two things:

    1) Reporting on a long weekend friday is what people do to bury news, reporting on a wednesday is what people do to maximize news

    2) Reporting on a fixed schedule would remove perceptions of bias going forward

    I see nothing wrong with either point. Both are 100% correct.

  34. This poll should scare the s#*t out of the tories-not because of how much they have dropped, but because of what could happen once an election is called. They should beware of the voliteility of the electorate , and how a small shift can make the 20 seat difference between a strong minority and losing government. This poll seems to verify the theory that EKOS somehow may have pushed the tory #s up to set them up for a fall, except that Angus Reid and Nanos both had similar big gains/losses for the tories and libs. Maybe Eric is right about people getting scared @ the thought of a Harper majority especially with the attack adds and Harper's support of Bev Oda despite her (not) lying and (not)covering up her role in the Kaios scandal. Once again, this is sand in the tory machine's engine. It will be interesting to see if this change is reflected in other polls or its just an "outlier"

  35. Conservatives widen lead to 15%:

  36. Eric RW there was also Harris Decima which only showed a +1 CPC gain and a -1 LPC loss during that period.

    So not all pollsters picked up on this huge volatile gain and then huge loss that EKOS is suggesting.

    Your point about the unpredictability of an election is a good one though.

    Right now the signs point to a CPC razor thin majority or just shy of that.

    But a couple "scandals" during an election can cause a lot of trouble.

    Just ask Ralph Goodale.

  37. There is also a CROP poll out for Quebec this morning.


    on francophones....

    And while 61% favor a coalition to keep Harper out.... only 6% of people think Iggy is the best choice for PM.... 4th to Harper/Duceppe at 16% and Layton at 17%

  38. I do understand about how you feel about conspiracy theories Eric.

    But they are easier to dismiss when things are done above board, and on a schedule.

    So ekos moved a big tory drop this week to wednesday.. a better news day.

    2 weeks ago they moved it to Friday when the tories did well.

    2 weeks before that they released on the Thursday schedule.

    They have bounced in recent weeks from 2800 respondents this week, down to 1600 two weeks ago, 4600, 3500, 2500 before that.

    They seem to have altered their reporting system several times, today they include only the last week in the 2 weeks, where they used to include 2 sets of polling in each release.

    Several pollsters participated in a discussion about how valuable polls are just 2 weeks ago... when.. oh yeah... the tories shot up in the polls.

    And Graves has been accused (more than once) of giving advice, money and support to the liberals.

    And 2 weeks ago (knowing the approximate schedules of other pollsters) ekos released a poll that matched everyone else. Then showed a precipitous drop this week.

    (which he blamed on oda.... while both HD and Abacus came ot with polls that noone knows who she is or what she had done)

    But Abacus and CROP don't seem to agree with his findings this week. A couple of firms, one new, one not on a schedule that might not have been included in a calculation of when others might come out with polls.

    So you can see how a conspiracy theorist might come up with a conspiracy theory on this one. Lots of them are based on less evidence.

    Then there was "computer problems" a few weeks ago where they lost some data.

    In the end, while share your "enthusiasm" about conspiracy theories. I am not sure I can believe the swing up and then down in the last couple weeks... nothing really having happened that could affect opinions.

    Does anyone want to bet that the next ekos poll (in 2 weeks on the week that several other firms come out) shows the tories over 36, and the liberals under 26? .... another swing. What is the analysis gonna say?

  39. And Elections Canada has charged the CPC and four of it's top people with fraud and money laundering.

    Along with the Oda odor that has to hurt !!

  40. "And Elections Canada has charged the CPC and four of it's top people with fraud and money laundering."

    Despite the fact that they have to win an appeal next month or some (alot) of the charges get dropped.

    When's the next deadline for liberal leadership debts from 07? December '11? How long will the next extension be for a debt that was supposed to be paid within months?? Maybe Elections Canada will offer up money for them... like their Manitoba cousins did for the provincial NDP

  41. Ira - so you prefer a system where a party with 35% support could potentially have 100% of the power to one where you need 50.1% support? Quite bizarre imo.

    I was here in Ontario when the NDP gained majority control for 5 years with 37.6%. The Chretien years had 'majorities' of 41.2%, 38.5%, and 40.9%.

    Now, some think that is the only way to get things done. The NDP one brought crazy deficits. The Chretien years brought in drastic cuts to health care & education, shifting deficits to the provinces who in turn shifted a lot onto cities.

    The last time we had 50% + 1 for one party federally was 1984. I think minority governments tend to be more responsive to the population, but to each their own.

    Hmm... an interesting project for Eric - what is the minimum vote % needed for each party to gain majority control (assuming all others drop at an equal pace)?

  42. Ira - so you prefer a system where a party with 35% support could potentially have 100% of the power to one where you need 50.1% support? Quite bizarre imo.

    ..... Probably not,... For me, I prefer a little stability, a little responsibility to the constant campaigning and vote buying in a minority.

    Unless you think Harper has been responsive to your green agenda to date... that Trudeau was responsive to the people when he instituted the NEP or when he gave the west the finger...??

    Tell me, which minority was more responsible with money?? The Harper minority where he was forced into a 50B deficit? the Trudeau minority which left us with a deficit in the billions and more social programs like healthcare that pull billions more of budget dollars to be spent inefficiently?? Or the Ontario example,.. where the '85 election accord with Bob Rae began the crazy deficits that followed years later under Rae's NDP government??

    An interesting system you support.... "Quite bizarre imo."

  43. Peter,

    It kind of undermines the credibility of your point when you get the facts terribly wrong. It's almost as if you haven't read any of the many stories on the issue. No one has been charged with fraud or money laundering. They've been charged with "willfully" exceeding the spending limit (presumably under subsection 497(3) of the Elections Act). Now, Elections Canada has already lost one round in the courts on a related point (which has been appealed by Elections Canada) so it's far from clear that Elections Canada has much of a chance of getting a conviction ( particularly given the necessity of proving that the violation of the election limit was "willful" which is typically the highest mens rea barrier for the crown).

  44. John,

    Here's the question I think anyone who wants to reform our voting system has to ask. What exactly is the problem that's so significant that it has to be resolved?

    Our system has worked well for the better part of 150 years (no small achievement - there aren't many other democracies who can boast that sort of record). Yes, from time to time we get the odd "strange" result (i.e., the NDP in Ontario in 1990, or the 1998 election in Quebec), but other voting systems produce equally "strange" (and often far more sinister) results (think of some of the dysfunctional coalitions in Israel, Jorg Heider's "Freedom" party in Austria, or the outcome of proportional representation in Weimar Germany).

    Democracy isn't an end in itself, it's a means to the end of producing good government. 150 years in, I'd say the Canadian democratic system has done pretty well for itself. Coming up with a new system is an interesting exercise in political theory, but I'm sort of attracted to the "if it ain't broke, don't fix it" school of thought.

  45. Very typical Carl. Spew incorrect and then call the other on them.

    First case was appealed by the Tories and their appeal succeeded at that court level. Elections Canada appealed that result and that case is NOTover yet as their appeal is yet to be heard.

    Second case has just started, it's not a criminal case so criminal law does NOT apply.

    Nice try, suggest you check your facts first !!


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.