Tuesday, April 10, 2012

March 2012 Federal Poll Averages

Due to the NDP leadership convention and the federal budget, a large amount of polling was done during the month of March. It is a difficult month to analyze, however, as the most important events took place in the final week. Though an incredible 16 federal polls surveying a total of 25,204 Canadians were released during the month, only a small handful (including two national polls) were conducted entirely after the NDP convention and before the month of April. This means that only a very small proportion of the "Mulcair bounce" is included in this monthly average.
Nevertheless, through the month of March the Conservatives averaged 34.9% support, a gain of 1.7 points compared to their February average. The New Democrats averaged 29.2%, a gain of 1.1 points, while the Liberals dropped 2.2 points to 21.9% support.

The Bloc Québécois gained 1.1 points to reach 6.9% nationally, while the Greens were down 0.6 points to 5.7%.

For the Conservatives and NDP this represents very little change. The Tories have averaged between 33% and 36% for the last five months, while the NDP has been between 28% and 30% for the last six. This is a bit of a drop for the Liberals, who have not been this low since September 2011 (when the NDP leadership race began). The last polls of March and the most recent surveys from April, however, indicate that the NDP's numbers have jumped even more, primarily at the expense of the Liberals.

The most significant amount of change came in Quebec, and we can already expect even more of a shift in the month of April. But throughout March, the New Democrats averaged 30.2% support in Quebec, a gain of 3.7 points and a return to where they stood in January. In fact, this is the first month of gain for the NDP in Quebec since September. The Bloc Québécois picked up 4.1 points and sat at 28.9% in March, their highest since the 2011 election campaign and their fifth consecutive month of growth. The Liberals slipped 3.2 points to 20% while the Conservatives also dropped that amount of support. They averaged 15.5% in Quebec in March, their lowest score since November 2010. The Greens were up 0.1 point to 3.9%.

Alberta also saw a good deal of variation, with the Conservatives picking up 3.7 points and averaging 57.2% support. The New Democrats were down 1.4 points to 19.1% while the Liberals were down 1.7 points to 12.5%. The Greens took the biggest hit, slipping 1.8 points to 5.5%.

The largest NDP regional gain in March took place in British Columbia, where the party jumped 4.1 points to 36.8% support. That is their highest mark since August 2011, the last time they led the Conservatives in a monthly average. The Tories were down 0.3 points to 35.5%, generally where they have been for the last five months. The Liberals were down three points to 17.2% while the Greens were down 0.4 points to 8.7%.

In Saskatchewan and Manitoba, the Conservatives were up 3.7 points in March to 43.7% support, well ahead of the New Democrats who averaged 34%, a drop of one point. The Liberals were down 2.3 points to 15.9% and the Greens were down 0.1 point to 5.8%.

The Conservatives picked up 2.4 points in Atlantic Canada to lead with 33.3%, while the New Democrats were up 0.9 points to 32.7%. The Liberals slipped 2.7 points to 26.6%, while the Greens were down 0.9 points to 4.9%.

The least amount of change took place in Ontario, where no one gained or lost more than 1.6 points. The Conservatives gained 1.2 points and averaged 37.8% in the province. They have been between 36% and 39% for the last five months in Ontario. The Liberals averaged 28% support, down 1.6 points. That is their lowest average result since September. The New Democrats were up 1.3 points to 26.8%, their highest result since that month. The Greens were down 0.2 points to 5.8%.

With these numbers, the Conservatives would have won 144 seats in an election held in March. That is an increase of 11 seats from the February projection. The New Democrats win 78 seats, an increase of four, while the Liberals drop 17 seats to 58. The Bloc Québécois wins 27 seats (+1) while the Greens retain their one.

The Conservatives win 18 seats in British Columbia, 27 in Alberta, 19 in the Prairies, 57 in Ontario, eight in Quebec, 14 in Atlantic Canada, and one in the north. In a 338-seat House, the Tories would likely have won 161 seats.

The New Democrats win 13 seats in British Columbia, one in Alberta, six in the Prairies, 24 in Ontario, 27 in Quebec, six in Atlantic Canada, and one in the north. They would take 84 seats in the expanded Commons.

The Liberals win four seats in British Columbia, three in the Prairies, 25 in Ontario, 13 in Quebec, 12 in Atlantic Canada, and one in the north. That would be bumped up to 64 seats on the new electoral map.

Considering that the New Democrats did not have a leader during most of the month and in the majority of these polls, there are already some good signs for the NDP here. They gained traction nationally in March and particularly in British Columbia, Quebec, and Ontario. The party was beginning to turn the corner in Quebec with the expectation that Thomas Mulcair would become the next leader, and recent polls show that the NDP might average more than 40% support in the province in April if things continue.

The Conservatives were holding generally steady in March, but bled a lot of support in Quebec and seems unable to keep the New Democrats down in British Columbia, a good place for NDP growth come 2015. But April is setting up to be worse for the government, as it is for the Liberals. March already indicated that the party is in trouble, as it took big steps backwards in Quebec and British Columbia and lost support in every other region. If the Liberals continue to place third in Atlantic Canada and lose second place to the NDP in Ontario, things will turn very sour for the party very quickly.

A last snapshot, then, of the political landscape in Canada before the arrival of Thomas Mulcair.


  1. I know this is the March averages, but when will you examine the poll that Leger Marketing released over the weekend showing the NDP and Conservatives in a statistical tie? With the NDP at 33% and the Conservatives at 32%?

    1. I have to prioritize - the Alberta election on Apr. 23, for instance, has precedence over the federal election in 2015.

    2. 1 in 20 outlier10 April, 2012 15:04

      The leger poll of April 28, 2011 had the Cpc in a near statistical tie. this was 3 days before Canada voted.

      A leger poll statistical tie would likely (due to safe CPC seats being created) a larger CPC majority.

      That is is the vote takes place today... not in over 3 years.

    3. 1 in 20 outlier, you seem to think we are all idiots. Go spew your talking points somewhere else.

    4. 1 in 20:

      Why is a year old poll so important?

      Obviously, the poll was not accurate since the Tories won by over 9%!

      It seems to me that even if Leger tends to favour Conservatives the probability of a larger or smaller Conservative caucus hinges more on the distribution of the vote than the polling firm.

    5. 1 in 20 outlier11 April, 2012 11:10

      So all the polls were wrong... Not one had some how over-estimated the CPC support as would happen if it was able to capture reality.

      This is the second federal election in a row that the polls were statistically wrong in predicting the outcome. Almost all the polls just prior to election night were the 1 in 20 outliers.

      With this knowledge gained from history how wise it suspend logical reasoning and scepticism and to accept these latest polls as an accurate representation of the polling universe?

      There is obviously something wrong in the pollsters methodology in that they consistently, 100% of the time find the CPC support lower than what it turned out to be.

      A fair and unbiased methodology would have had some outliers on the other side of the curve... there should be 50% of the samples having the CPC at above 40%...... but there was not even one.

      Eric used to track pollster bias based on how far they were off from the "average" poll. He has dropped that as raising that question of bias leads directly to the question of how biased are the polls to the actual results....... and this makes all the pollster look bad.

      These pollsters are well respected by the MSM and are making a good living as commentators/pundits/personalties. They have been getting away with this despite them consistently being inaccurate.

      If Eric were to incorporate historical polling error/bias into his seat predictions the pollsters would stop returning his calls and e-mails.

      He would be more accurate and become the go to polling analysis expert BUT would be outside the loop that he is just getting into.

      Eric, it seems is more comfortable, being way way off in his seat predictions and doing a follow-up explaining that his model would have been bang on had only the pollster been somewhat accurate.... Then immediately , the next day, accepts the same pollsters info as being accurate and requiring no adjustment.

    6. My current Alberta projection makes adjustments based on past discrepancies, as my future projections will.

      But when it comes to polls, I report on what they actually say. I'm not going to make adjustments here and there based on one or two data points from 2008 and 2011 every single time. If those pollsters in 2011 had done their polling 10 times on the final day, they would have had 10 different results. That is just how it works, so it is wrong to simply assign bonus points to individual polls based on what happened on Apr. 29, 2011, or whatever.

      Polls track public opinion, I report public opinion when talking about the polls. When 2015 rolls around, I will be making adjustments to my own projections, but that doesn't mean I will assume every individual poll is wrong and is going to be wrong in the exact same way that it was in 2011 - that would be wrong. One or two elections does not make a trend that can be expected to be repeated again and again in future elections. We're looking at individual cases, maybe even exceptions.

      I stopped tracking pollster house effects because they were misleading. Comparing polls taken on different days that the pollsters make public (many firms are constantly in the field, but they don't release their data every week) was not a good way to go about it. And you must have missed my posts and articles after the 2011 federal and provincial elections, when I compared the results to what the polls were saying. Their errors (and successes) were laid bare for everyone to see.

      You're wrong that pollsters are "consistently" inaccurate. If you have so low of an opinion of their work and mine, I suggest you stop visiting the site.

    7. First, some May 1 polls were within margin of error of the election results.

      Second, some pollsters have adjusted methods in light of the discrepancy between pre-election polls vs. election outcome.

      Third, your explanation of the discrepancy seems to be rooted in some assumption of systematic pollster bias against the CPC. Have you even followed the industry and academic debate on this question? Have you looked at this:


      I think you will find some answers there.

    8. 1 in 20 Outlier11 April, 2012 15:56

      Hi Eric

      If you don't get dissenting views you will lose whatever critical edge you currently possess.

      Thanks Jeff

      I read the ekos explanation and when someone starts

      "This exercise is not an attempt at apology orrationalisation."

      and ends with the conclusion that their poll results more accurately reflect the opinion of Canadians than the electorate:

      "In closing, these are very complex and difficult questions. The Conservative Party of Canada won
      a hard fought and legitimate majority mandate working within all of the rules of the current electoral system. There are however, profound issues looming about how to register and take into account the huge portions of Canadian society who are soon to be the mainstream and who have systematically opted out of the electoral process. This may be a far greater challenge to polling and democracy than the somewhat suspect polling sweepstakes of who came closest to the final vote outcome"

      What I get from Mr. Graves lecture is:

      Pollsters are not wrong the Canadian voting public is wrong and there is no real need to adjust the polls rather just get the people to vote properly.

      It just seems the pollster is biased and proud of it. If it was me paying for Mr. Graves polls I would still want a refund.

    9. There is no bias there, he is merely pointing out the difference in voting intentions of people who vote and people who don't. It isn't about "voting properly", merely voting.

    10. 1 in 20 outlier11 April, 2012 17:39

      A pollsters job is to provide a point in time reading on probable election outcomes.

      It is not to fix democracy or to make hypothetical cases of what would happen if the 5% of people actually answered the phone call would vote if someone could accept their vote on a phone conversation.

      The "why I got the election so wrong" essay did not provide information on how many voters were called and how many answered and how many hung up and how many would actually vote.

      Having a view on a problem and possible solutions is not what pollsters are supposed to be doing... Soon as they become part of the process they are inundated with bias.

      If the poll results are do not express a clear outcome or the underlying reasons for the same results on data points alone then it a badly constructed poll. The suppositions as to what drove the polls results should not be provided by the pollster.

      Almost all of the Canadian pollsters have become analysts and pundits. The temptation to design and interpret the data to suit their views and biases has won out.

    11. What Eric said.

      In other words, Monsieur Outlier, the central point of Graves' essay was in fact to openly admit error in projecting the election outcome but also:

      "disentangling the two
      separate challenges of: (i) forecasting the election result and (ii) modelling the population of all
      eligible voters. Having made an error on (i), we ask the reader not to confound it with the second

      Maybe the opinion of the non-voter means nothing to you, Outlier. But the political Right in North America pays close attention to that non-voter and has been intent on moving more and particular segments of the population into the camp of non-voter through general and specific voter suppression techniques.

      I personally would like to see more specific details on the changes pollsters have introduced to better project electoral outcomes. In the meantime, no one will be convinced by Outlier's self-serving argument that because a few polls modestly underestimated relative Conservative turnout in May 2011, that we should ignore the polls altogether and assume the CPC will be dominant for time eternal.

    12. 1 in 20 outlier12 April, 2012 11:47

      CPC will not be dominate for time eternal.

      Based on the historical polling results when we see the NDP or Liberals ahead by 4 points in the poll it will mean that they are even. When we see the NDP or Liberals ahead by 8 they will likely win a plurality of seats in an actual election.

      I do not believe that the COMPAS polls that had the CPC at 46% going into the election are any more accurate than the polls that had the CPC at 34%.

      The role that the pollsters have migrated to, punditry and political analysis, IMHO, cannot do otherwise than introduce bias.

      Are the COMPAS polls any less scientific rigorous than the other polls? Yet it can come up with results that are as equally biased and again in my opinion (and based on statistical evidence) wrong.

    13. 1 in 20 outlier12 April, 2012 11:51

      As for moving the non-voting young person to the right.... getting a job and paying taxes which is a function of aging accomplishes that.

      By the time the youth cohort that Graves moans as not being counted decides to be counted they will be voting much closer to the centre and centre right.

  2. Someone should invent a dance called the 'Mulcair bounce'. Perhaps it could involve rhythmic sleeve-rolling?

    1. And indignant glaring.

    2. Good one! All jokes aside, I think it needs to be said that there is no such thing as a Mulcair "bounce". Or a Mulcair "honeymoon". Or any other euphemism for the lurking sentiment: "uh oh NDP rising in the polls must be temporary".

      These descriptions represent subjective interpretations of an objective polling trend, and are without empirical foundation at this point in time. All bounces and honeymoons are presumed to come to an end. Will this NDP "surge" come to an end? Maybe yes, maybe no. And even if it does it won't be because of some law-like force of polling Nature. Did Stephen Harper's "bouncy honeymoon" come to an end? Or is that raunchy, squeeky affair still keeping us up at night?

      Above Eric observed that the NDP has held most of its support in recent months, this despite being without a permanent or effective leader and much of its front bench. This seems to be a clear empirical refutation of the last round of euphemism passed off as explanation by the lazy punditocracy: remember the ephemeral blip they dismissed as "Jack-o-Mania" and a passing "Orange Wave"?

      The fact is we have no idea what will happen, partly because there are no irrational or irresistable forces of romance or gravity that are drive polling trends. Real people and events will determine the electoral fortunes of the political parties.

      P.S. Personally, I think the NDP is merely coming out of its "Turmel trough".

  3. So Eric, any news about Alberta election polls?

    1. +1.
      Love your site Eric, and I check back compulsively.
      Look forward to more on the Alberta Election.
      New poll reported by the Herald has WR and PCs neck in neck (too bad for me)

      Go Wildrose!
      40 Years is Long Enough!
      Kick the Bums out!

      Wildrose Supporter (Former PC Supporter)

    2. The projection was updated yesterday afternoon with the new Léger poll.


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.