Friday, March 11, 2011

No major shifts in new Angus-Reid and EKOS polls

Yesterday, we were treated to two new polls, one from Angus-Reid and the other from EKOS Research. They told somewhat different stories, as Angus-Reid pegs the gap between the Conservatives and Liberals at 16 points, while EKOS has it at around seven.

How do we reconcile these two polls? We don't!

I'm not going to compare the two polls as they aren't comparable. Angus-Reid uses an online panel, while EKOS uses a telephone system. Angus-Reid polled on two days (March 8 and 9), while EKOS polled on nine days (February 24 to March 8, excluding weekends). Finally, Angus-Reid polled 1,021 people, while EKOS polled 2,892 people. The two polls aren't at all the same.

But what the two polls do have in common is that neither shows a significant shift in support for any of the parties since the last time these pollsters were in the field.

We'll start with Angus-Reid, as it is the most recent poll of the two.

Compared to Angus-Reid's last poll (February 11 to 18), the Conservatives have remained unchanged at 39% support. The Liberals are down three points to 23%, while the New Democrats are down one to 17%.

The Bloc Québécois is steady at 9% nationally, while the Greens are up three points to 9%.

Angus-Reid reports a margin of error of +/- 3.1%, 19 times out of 20. They should be reporting that the sampling margin of error of a random sample of similar size is +/- 3.1%, or not reporting an MOE at all.

But if we take this as a random sample, we can say that both the Liberal and NDP drops are not statistically significant. The Green jump is, but just barely.

In Ontario, the Conservatives are down two points to 41%, followed by the Liberals at 29% (-1). The NDP is steady at 19%. These are horrid numbers for the Liberals.

The Bloc has dropped five points in Quebec, and now leads with only 34%. The Liberals are up two to 23%, the Conservatives are up two to 22%, and the NDP is up one to 15%. Generally steady-as-she-goes here.

In British Columbia, the Conservatives are up five points and now lead with 45%, while the NDP (-10) and Liberals (-4) are tied at 18%, very low results. The Greens are up five to 13%.

In Alberta, the Conservatives are up nine points to 69%, while the Liberals are down 12 to 9%, joining the NDP. And in the Prairies, the Tories are up two points to 57%, the NDP is steady at 22%, and the Liberals are down four to only 15%.

This is wasted support for the Conservatives. They can't make any major gains in Alberta, Saskatchewan, and Manitoba, so all they are really doing is winning the seats they already have by a lot more.

Atlantic Canada is a write-off in this poll. The Conservatives are down 12 points to 27%, the Liberals are down 14 to 26%, and the NDP is up four to 21%. The Others are up 19 points to 20%, which is of course ridiculous. This isn't the first time that Angus-Reid has had the Others up to an implausible level of support in Atlantic Canada, which tells me they may have a few Maritime jokers on their panel.

This doesn't break the poll, though, as that 20% only accounts for about 1.4 points nationally. But Angus-Reid probably should have extended their polling period out East to try to correct the error.

With the results of this poll only (and ignoring the Others in Atlantic Canada), ThreeHundredEight's projection model gives the Conservatives 26 seats in British Columbia, 28 in Alberta, 22 in the Prairies, 57 in Ontario, 11 in Quebec, and 13 in Atlantic Canada for a total of 159. That's a majority.

The Liberals win four seats in British Columbia, none in Alberta, two in the Prairies, 32 in Ontario, 17 in Quebec, and 14 in Atlantic Canada for a total of 70 seats. That's a disaster.

The Bloc Québécois wins 44 seats in Quebec, quite a bit of a drop for them.

The New Democrats win six seats in British Columbia, none in Alberta, four in the Prairies, 17 in Ontario, two in Quebec, and five in Atlantic Canada for a total of 34.

André Arthur manages to barely hold on to his riding of Portneuf - Jacques-Cartier.

Angus-Reid also looked at how Canadians see the major party leaders. Stephen Harper has an approval rating of 32% and a disapproval rating of 45%. Compared to Angus-Reid's last poll which involved that question, the approval/disapproval spread has improved by 11 points.

For Michael Ignatieff, he has an approval rating of 14% and a disapproval rating of 55%, a spread that has gotten worse by eight points.

And for Jack Layton, his approval rating is 34%, compared to a disapproval rating of 30%. That's a spread that has improved by 13 points.

Angus-Reid also looked into how Canadians viewed their qualities and flaws. Looking at only the characteristics that have seen major shifts since Angus-Reid's last similar poll, fewer people (+4% each) think Harper is secretive, efficient, and out of touch.

More people think Ignatieff is arrogant (+6), uncaring (+5), and dishonest (+11). Ouch. But fewer people think he is boring (-5).

And for Layton, fewer people (-4 each) think he is compassionate, exciting, and intelligent. Not exactly terrific.

Alright, on to EKOS.

Compared to their last poll (February 10 to 22), the Conservatives are up 2.8 points to 35.2%. The Liberals are up 0.5 points to 27.8%, while the NDP is up 0.1 points to 14.9%.

The Greens are down 1.8 points to 10.1% and the Bloc Québécois is down 1.7 points to 8.8%.

This poll has a sampling margin of error of +/- 1.8%, 19 times out of 20. About 14% of Canadians are undecided (a drop of about three points). And none of these national shifts in voting intentions are statistically significant.

The Conservatives are up 5.1 points in Ontario and now lead with 41%, a very good number for them, but not unusual of late. The Liberals are down 2.4 points to 34%, while the NDP is up 0.2 points to 14.4%. The Conservatives are leading in Toronto and Ottawa with 41.1% and 44.8%, while the Liberals are running second with 38.8% and 36.6%, respectively.

In Quebec, the Bloc is down four points to 35.9%. The Liberals are up 2.9 points to 21.7%, while the Conservatives are down 0.8 points to 15.4%. The NDP is up 1.1 points to 12.5%. The Bloc leads in Montreal with 38.5%, followed by the Liberals at 26%. The 10.1% score for the Conservatives is not good for their West Island hopes.

The Conservatives are up 3.3 points in British Columbia and now lead with 36.2%. The Liberals follow with 24.1% (+3.1), while the NDP is down 3.3 points to 21.9%. The Greens are also down, dropping 5.2 points to 14.2%. The Conservatives are leading in Vancouver with 39.2%, followed by the Liberals at 26.6%.

In Alberta, the Conservatives are up 5.2 points to 57.3%, and are trailed by the Liberals at 17.2% (-7.2) and the Greens at 11.3%. The Conservatives are leading in Calgary with 60.7%, with the Liberals at 17.4%.

In Atlantic Canada, the Liberals are up 9.8 points to 40% and lead once again in the region. The Conservatives are down 5.9 points to 30.9%, while the NDP is up 3.1 points to 17.4%.

And in the Prairies, the Conservatives are up 3.4 points to 45.7%, followed by the Liberals at 22.5% (+0.9) and the NDP at 21.1% (-6.7).

Again, with the results of this poll only, ThreeHundredEight's projection model gives the Conservatives 21 seats in British Columbia, 28 in Alberta, 21 in the Prairies, 56 in Ontario, eight in Quebec, and eight in Atlantic Canada for a total of 144. That's a minority.

The Liberals win seven seats in British Columbia, none in Alberta, three in the Prairies, 35 in Ontario, 17 in Quebec, and 20 in Atlantic Canada for a total of 83.

The Bloc Québécois wins 49 seats in Quebec.

The New Democrats win seven seats in British Columbia, none in Alberta, four in the Prairies, 15 in Ontario, one in Quebec, and four in Atlantic Canada for a total of 31.

The Greens manage to get Elizabeth May elected in Saanich - Gulf Islands.

EKOS also took a look at second choice, which hasn't changed much from their other polls. But they also looked at how people's vote has shifted since 2008, which is very interesting.

The Conservatives are the best at keeping their vote, as about 77% of their supporters in 2008 are still intending to vote for them. Another 12%, however, have went over to the Liberals.

The Bloc is the next best at keeping their vote, with 74%. About 8% has gone to the Greens and another 7% has gone to the NDP.

The Liberals are maintaining 64% of their vote, but have lost 18% to the Conservatives.

And the New Democrats have kept 62% of their vote, with 13% going over to the Liberals and 12% going over to the Conservatives.

This tends to show that anywhere from 20% to 40% of a party's supporters are liable to head elsewhere from one election to the next. That's a lot of movement, and not a lot of it is homogenous. You have voters going every which way.

Both of these polls are good for the Conservatives. Angus-Reid shows them with a huge lead, but more importantly both Angus-Reid and EKOS have the Conservatives at 41% in Ontario. This is where the Conservatives intend to win their majority. If they can maintain this sort of support in Ontario, and bring back their voters in British Columbia and Quebec, they will be in a very strong position come election day.


  1. I was thinking about online polls last night, while reading the comments section at The Globe's website.
    How hard would it be for all the extreme partisans that infect the comments section of most news sites to game an online poll by signing up multiple times, lie about their previous votes, and generally mess up the results?
    The more I think about these people (who seem to either unlimited time on their hands to rant about politics, or are paid shills), the more I come to completely write off the results of any online poll that I see.

  2. Excellent results for the CPC. I would like to see the last couple of days polling for EOKS to see if any of the mud being thrown is sticking.

    The biggest takeaway is that Iggy is not really considered an alternative.

  3. One thing I find curious in the Ekos poll is that about 50% of people who claim to have voted Green in the 2008 election say they would now vote for some other party and only small dribs and drabs of people who say they voted Liberal, Tory or NDP in '08 say they would now vote Green. Given that in the '08 election the Green party got 6.8% - that SHOULD take them down to 3.4% for starters with bits and pieces of migration from elsewhere plus a few people who would have turned 18 in the past two years - but I see no way that you can get them to over 10% as Ekos says. I would be REALLY curious to see the results of the question Ekos must have asked "Who did you vote for in the last election in 2008?" and see if wayyyy more people claim to have voted Green than we know actually did.

  4. I'm wondering if Iggy & the Liberals have decided to just sit back, let all the mud fly from the CPC (since they really can't afford to fight back pre-election) then jump in as hard as possible once an election is called, hoping that the average voter will decide Iggy isn't as bad as stated (low expectations) and thus be able to regain lost ground quickly and before the CPC can dream up a new wave of anti-Iggy ads.

    Just trying to think through what I'd do if I was a Liberal (ugh) or advising them. Probably sitting back and letting as much mud be flung as possible, then trying to look like the opposite of the characterization would be the best advise possible. No idea how they can gain enough though to win.

    If the Liberals drop to 20-25% nationally though on election day boy will it be weird and ugly. An NDP/Liberal merger would be almost automatic as both would see it as the only hope, with an attempt to draw Green's in as well and shift Canada to a 2 party system (outside of Quebec).

  5. Eric, what if any significance do you give to the fact that Angus-Reid selects their sample from members of a pre-constituted "forum"? (From the report: "From March 8 to March 9, 2011, Vision Critical/Angus Reid conducted an online survey among 1,021 randomly selected Canadian adults who are Angus Reid Forum panelists.")It seems to me that the kinds of people who sign up for this sort of thing -- like the partisan posters Dan F complains about -- are unlikely to be representative of the general population, if only because they are more interested in politics than most.

  6. Anonymous,

    The use of online panels is a controversy and a big issue in the polling industry, so there is little I can add to the debate.

    But remember that Angus-Reid is not just a political poll firm, they poll on a myriad of issues for private companies and other organizations.

    I used to be a panelist myself, and most of the surveys had to do with commercial products.

    So, the panel isn't made up of die-hard partisans. I don't think the particular concern you highlight is a likely problem.

  7. Dan F

    Really really hard.

    I have signed up for the AR vision critical universe about 4-5 months ago to better see the inner mechanisms.

    I have done 4 getting to know you surveys which basically is establishing my demographics.

    There has not been a hint of a political question other than how did you vote last time and have you ever voted for or been a member of a party. (The only time I was a party member was when a friends dad was running and he bought me a Liberal Card)

    I can see they have enough information to skew the sample to get whatever result they want but as they are the most accurate by long shot they must be doing something right.

  8. DAN...yes it is possible that anyone person could vote multiple times on a standard on-line poll (for example, a polling question of "which party will you vote for in the next election?")that appears on an internet home page (eg msn, bell sympatico etc...) this case anybody that does that is just fooling themselves...i for one refuse to vote more then once and also,usually so many people respond to these polls that a few delinquents won't make much of a far as formal on-line polls being conducted by,for example, angus-reid, i don't think anybody can vote more then once as the polling site "system" will recognize the i/p adress and disallow a second entry.there are ways of "messing around" no doubt but again, anybody that would go out of their way like that would be few and again wouldn't really skew the results....(lets also remember that angus-reid is one of the more/most accurate polling firms...)

  9. Some things I found interesting in those poll releases:

    In the AR poll, Jack Layton's positive approval rating marks the first time in six months that any of the federal leaders has had a positive approval rating. Also, that "use six adjectives to describe the leader" question is awful for Ignatieff. That AR releases one of these every couple of weeks can't be good for the Liberals.

    And in the Ekos poll, the analysis in the release that shows how Canadians, despite claiming that they would move left if they couldn't vote fortheir first choice, actually tend to move right when they do move. It seems that Canadians, like much of the world, still views Canada as a country that values socialism like a warm blanket. Even though it's not really true.

    There was an article in The Economist last month which went on at length about how Canada is "proud of its welfare state" and has its "big government intact" - bit it doesn't. In the last 15 years, Canada's government has shrunk from 22% above the OECD average to just below the OECD average. Government spending, which used to be 53% of CDP, is now only 40% of GDP (by comparison, US government spending is 39% of GDP - current trends suggest they'll pass us next year).

    Ekos is showing us that even though Canadians no longer behave as if they want a safety net and bigger government, they still say they do when asked.

    It's an interesting phenomenon.

  10. On the MoE issue, yes, AR should be reporting the MoE differently. But so should every pollster. There's a lack of randomness in online polls, but so is there in phone polling.

    You can't really get a random sample unless you can choose from among all Canadians, and you can force perticipation. Without that, there's guaranteed to be a selection bias.

  11. Don't forget that traditional (phone-based) polling has substantial biases of it's own. I was surprised to hear that the success rate (percentage of people who were phoned but declined to participate in a poll) is around 15%. I am sure being more partisan increases the chances to agree to participate in a phone poll, and this is cannot be easily corrected (unbiased) in the analysis. You end up sampling relatively extreme wings of the political spectrum. and this type of sampling has a very poor convergence to a "true mean" when you increase the size of the sample. (So the error should be significantly larger than purely random MOE).

  12. and right after I mention "extreme partisans that infect comments" Earl appears. Funny coincidence.

    Regarding the polls that just came out; the phone poll shows completely different results than the online poll (possibly even outside the MoE?), so clearly one of them is wrong.

  13. dan F.

    History shows that AR is the most accurate pollster and EKOS is in the also ran pack.

    EKOS seems to get the Liberal support very well but has the Green as 4-5 pts too high and CPC 3-4 points too low.

    There is no rationale too this , but that is what happened in 2008.

    Eric will tell you that the AR ranges at the bottom of the MOE is close to the EKOS top of the range MOE..... but it is consistent poll after poll

  14. EKOS has a good track record, and is my third-highest ranked national pollster. I wouldn't call them also-ran.

  15. Have you looked at the accuracy of the pollsters in 2004 and 2006 or just 2008?

  16. I've looked at the records of the pollsters going back to 2008 but incorporating all provincial campaigns as well.

  17. Any online poll or question of the day I always check to see if one can vote more than once from the same computer. You get, we have already counted your vote-those I believe. Those that never question if you have voted, I don't believe them.
    Then there is some way one can do something with cookies that allow you to vote more than once. Found that out when the UFA, in Alberta, had a Vote for your local hero. You could vote once a day, and all of a sudden an new nominee was receiving thousands of votes/day. I reported it, and they then had to have you enter some code word.
    The prize was a concert om tje winners home town, and the President of UFA, in his speech said x number of votes had been cast in total for all candidates, and most of them were legal.
    And there is no way saying you voted for x last time and would vote y next time can be proved.

  18. Angus Reid only polled in 2008 so it's possible they're not the most accurate. Nanos' last poll in 2006 was 0.1% off for all of the parties except for the Greens.

  19. Wow. Just look at those amazing blue towers.

  20. Nanos's last poll in 2008 was off by more than 3.5 points on Tory support.

  21. Dio,
    Which means Angus could be off by that much or more.

  22. Eric

    "So, the panel isn't made up of die-hard partisans. I don't think the particular concern you highlight is a likely problem."

    I've been on AR's panels for over five years but what I've noticed is that consistently voting Liberal means I see fewer and fewer polls.

    For instance this last one you quoted was NEVER offered to me !! So no bias, eh ??

  23. This article seems to suggest that it's those incessant attack ads that are driving down Ignatieff:
    And the sad thing, for the Liberals, is that they aren't good enough fundraisers anymore to mount a response. Ignatieff's like a political pinata. A Chretien would have gone out and given some fiery speeches defending himself that might have resonated with the public, but one don't even see Ignatieff's counter responses making much headway. He is being defined, negatively, month after month, by the Harper team.

  24. The Nanos poll of Friday Oct 10th 2008 had the CP at 32% and the Liberals at 28%.

    The campaign was for all intents and purposes over and Nanos called the election as a statistical tie.
    On Oct 14 (election day) CPC 37.6 Liberals 26.2.

    If Nanos actually believed his poll results of Oct 10,2008 .... and he did enough to publish them. He should be writing a couple of books on how the CPC used the last weekend to gain 5.6 points. It was the greatest political weekend in Canadian history.

    OR you can look at AR who had the CPC at 38% on Oct 10 and The Liberals at 28%.

    On Oct 10,2008 EKOS and Harris Decima both had the Green at 11% rather than the 6.8% they got

  25. Peter

    Maybe you and I have been identified by AR as super partisan... I have not been invited to any political polls.

    Or Maybe your name doesn't come up at a winner out of their 100,000s on the panel.

    Much the same as you can lose a lottery.

    I believe that they are still doing a random sample on their universe which they consider to reflect Canada.

    I one I have trouble getting my head around is EKOS significantly over-polling in Quebec. It is almost as though they can't get anyone in the West to talk to them. That really throws off all sampling theory.

  26. The Quebec sample in EKOS's last poll was larger, but mostly at the expense of Ontario, not the West. It could just be that they thought the Quebec results were showing something different, and so polled more there to make sure they had it right. It would've been weighed correctly afterward.

    As to Angus-Reid, it seems extremely unlikely they would not select participants because they've deemed them partisan Liberal or Conservative supporters.

    Far more likely is that the invitations to the panels is randomized so that it isn't always the same people answering them.

  27. Eric EKOS over-polls in Quebec consistently.... If they are trying to find something special in Quebec that requires a larger sample they are still looking.

    31 % of their poll for 23% for the population. or 24% of the seats

    You have pointed out that a lot of other pollsters do as well.

    Hard to figure.

    Maybe it is why Mr. Gregg came out and basically said that all Canadian political polls were of limited-doubtful value.

  28. Shawn

    I humbly suggest that is not the CPC attack ads that have hurt the perception of Mr. Ignatieff. Of course they haven't helped.

    I suggest that it is the decorum and strategy of his party in parliament.

    They have been openly following the rat pack strategy (Copps, Tobin, Boudria, Nuniziata) that was acclaimed as worked in 1984-93. It is the Liberal blue book for behavior in opposition.

    It has been credited with the Mulroney/Campbell trashing in 1993 but I say that the Bloc and Reform break away had very much more to do with it than boorish behavior.

    In 2008-11 there is a lot more media than in 1993. Actually seeing the committees at work creating their Liberal attack ads is back firing.

    All the negative attack ads do not have the blow back of 2 years of living a poorly constructed attack ad in Parliament as the Liberals have done.

    Almost all Canadians do not see Canadian democracy being threatened. They realize we have one of the best most open free democracies that is the envy of 90% of the world population.

    The Liberals say not.

  29. I get the AR poll and it always says, open to the first x number of respondents, by x date.
    And not all of them ask who you voted for or who you will vote for.
    But, re the rest of the questions, on most of them I have to mark never,
    on questions re, how often do you shop at x store, or how often do you drink starbucks, shop at costco or other outlet, or what financial institution you use, what communications do you use.
    Reason, not any of the products are available in my area. There is no place to put that answer, so I think it would skew the results.
    Same goes for question re public transportation, if it is not available you can't use it.
    I also refuse to answer any question re income, savings, investments. And your answers must add up to x total on certain questions, so when they asked how many glasses of wine, etc do you drink in a day, week, month, year, and had other beverages, listed, I had to answer 100 cups of coffee/day to get to the next question.

  30. BCVOR you could be on to something regarding a backlash against constant and unreasonable negativity directed at our prime minister.

    Here's a good example, Dmitri Soudas sends out this heart warming tweet:

    And Liberal candidate Jennifer Pollock reacts this way via twitter:

    "I agree with those who see a manipulative opportunist in this photo- #Escher #cdnpoli #lovethevideo"

  31. That fits very well into the context of this discussion (i.e., people seeing things that aren't there).

    It's too bad politicians don't allow themselves to show a little personality more often. I thought it was a nice thing for Harper to do.

  32. "Almost all Canadians do not see Canadian democracy being threatened. They realize we have one of the best most open free democracies that is the envy of 90% of the world population."

    This is blatant rubbish, Orwellian "double Speak". At any given time, 60-70% of the population will not support the Conservatives. Even more laughable given the egregious lack of transparency (now formally recognized in a contempt ruling) that the Harper Government(tm) has made their modus operandi from day 1 of their tenure.

  33. anonoymous

    Do you believe that you will not be given the right to vote as you see fit or to write into papers or forums with your opinions?

    Do you really think that the system is broken because you are for a brief moment in time not getting your concept of utopia implemented?

    The biggest group of Canadians are having their societal ambitions best represented by the CPC.

    How is this Orwellian?

    There definitely are no barriers to voting the CPC out of office. There will be no repercussions if the bulk of Canadians feel as you do that there is "egregious lack of transparency" by the Harper Government sufficient to banish them from public office.

    I have not heard of the rounding up of Liberal, or even BLOC dissidents being assigned to re-education camps on Baffin Island.

    Clear and simple Canada has a great democracy and this point can not be successfully argued even with Chretien (40% majorities) or even the NDP-Liberal-Bloc coalition in power that democracy has even been slightly threatened.

    Democracy would either validate this in the next election or banish them from power for a long long time. No one would expect them to even protest about the democratic will of the people.

    Canadian Democracy is not rubbish

  34. We have often debated constitutional matters around the issue of what happens after an election. Here is an excellent explanation:

  35. After this past week of the PC's shooting themselves in the foot numerous times, it should be interesting to see the next poll numbers.

  36. Does Tim Powers read this blog for ideas?

    Does he and the CPC election machine get their "talking points" from the grass roots?

    This weekend on "Question Period" he highlighted that if the Opposition continued using the committees as kangaroo courts it would hurt them.

    It would make no sense for Harper to call an early election when the Liberals and NDP were continuing to self destruct in committee hearings


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.