Monday, June 10, 2013

May 2013 federal polling averages

We were spoiled a little with 10 polls in April. In May, only four polls (three of them national, one of them conducted in Quebec only) were released - perhaps due to some shyness after the B.C. election. Nevertheless, some 5,000 people were surveyed. The sample is not as robust as it could be, but the results point to continuing positive trends for the Liberal Party under Justin Trudeau.
The Liberals averaged 40% in polls conducted in May, a gain of 6.6 points from their April averages. This is the highest Liberal number on record (going back to January 2009, as will be the case for the rest of this post) and the largest month-to-month gain by any party since the New Democrats jumped nine points between April and May 2011.

The Conservatives averaged 27.6% support, a drop of 2.6 points and the lowest number they have put up since at least the beginning of 2009. This is the fourth consecutive month of decline for the Conservatives, and the gap between themselves and the Liberals is the largest since the Tories had a 12-point lead over the Liberals in April 2011.

The New Democrats were down 0.5 points to 23.3%, their lowest since April 2011. They have been on the decline or stagnating for 11 consecutive months now (with the exception of a one-point uptick in January).

The Greens were down 1.5 points to 4.1%, putting them tied with the Bloc Québécois (down 1.4 points). Support for other parties averaged 1%.

Those are some pretty striking numbers, after a month of striking numbers in April. The trends all point to Liberal gains primarily at the expense of the Conservatives. EKOS, Forum, and Ipsos-Reid were all in the field in April, and the change from those polls is clear.
The New Democrats have shown a small gain from those April polls, however, in the unweighted average. But the April averages also included polls from other firms.

The Liberals made gains in all six regions of the country, while the Conservatives decreased in support from coast to coast.

In Ontario, the Liberals averaged 41.6% support, a gain of 5.8 points since April and their best result since April 2009. The Liberals have gained 16 points in Ontario since January. The Conservatives dropped 2.7 points to 32.6%, their lowest on record. The New Democrats were down 0.6 points to 21.7%, while the Greens were down 1.9 points to 3.4%.

The Liberals averaged 41% support in Quebec, their best on record. That represented a gain of 5.2 points since April and 17 points since March. The New Democrats had a small increase of 1.3 points to 27.7%, while the Bloc Québécois was down 4.6 points to 17.3%. That is their lowest number since October 2011. The Conservatives were down 0.7 points to 10.1%, their lowest on record, while the Greens were down 1.1 points to 3.1%.

In British Columbia, the Liberals picked up 5.4 points to lead with 32.9%, their best on record. The Conservatives were down 2.5 points to 29.1%, their worst on record, while the New Democrats at 28.2% (-2.4) scored their worst result since October 2011. The Greens were down 0.2 points to 8.9%.

The results in Alberta are probably anomalous due to an unusual result in Forum's polling. But nevertheless, the Conservatives fell 8.3 points to 47.2%, their lowest on record, while the Liberals were up 11.3 points to 33%, their best. While the size of the gain was probably inflated, the Liberals did have their fourth consecutive month of increase in Alberta. The New Democrats were down 0.3 points to 13.1%, while the Greens were down 1.5 points to 5.4%.

In Atlantic Canada, the Liberals put up the highest number any party has since at least January 2009 with 53%, a gain of 4.8 points. The New Democrats were down 4.3 points to 20.6%, their worst since March 2011, while the Conservatives were down 0.8 points to 20.5%. The Greens were up 0.6 points to 5.5%.

And in the Prairies, the Conservatives fell 3.1 points to 37.6%. The party has generally been on the decline here since January. The Liberals put up their best numbers with 33.4%, a gain of 5.5 points, while the NDP was down 3.7 points to 22.4%. That was their worst since March 2011.
With these levels of support, and on the 338-seat map, the Liberals would win 152 seats. The Conservatives would take 109 seats, the New Democrats would win 74, the Bloc would hold onto two, and the Greens would retain their one.

Compared to April, this is a 29-seat gain for the Liberals, a 22-seat loss for the Tories, and a seven-seat loss for the Bloc. The NDP and Greens were unchanged at the national level.

The seven seats in Alberta may seem unusual - and they are. But this is what happens when the Liberals are at 33% in the province. Will they actually get 33% in Alberta? Probably not, but this is what that number gives them. And, in any case, it doesn't change much if you give all seven of those back to the Conservatives.

The Liberals made their biggest seat gain in Ontario, jumping 13 seats from April - all at the expense of the Tories. When the two parties are neck and neck in the province, it is to the Conservatives' advantage. But with a nine-point lead, the Liberals are able to win the majority of Ontario's seats.

Quebec is still a puzzle. The Liberals have traditionally not put up good numbers outside of Montreal, and that was particularly the case in the last election. So when the Liberals triple their support in the province, they do not win a lot of seats outside of the Montreal area in the model when the base is 10% or less. This might inflate the NDP's potential seat wins - unless the party is capable of keeping the Liberals down below 30% in most of the province's majority-francophone ridings.

There were no new numbers on the Best Prime Minister question in May, but there were some approval ratings. Stephen Harper dipped below 30% for the first time, while Trudeau flirted with almost 50% approval.

The Trudeau honeymoon continues, undoubtedly aided a great deal by the plethora of problems the Conservatives are going through right now. If the Conservatives are to make it back, they have a lot of work to do. The last time they trailed in the polls - behind the NDP in mid-2012 - the party was still in a very strong position in Alberta, the Prairies, Ontario, and Atlantic Canada. They were even holding steady in Quebec. And the last time the Tories trailed the Liberals in the polls - in the spring of 2009 - they were still comfortably ahead in the West and within spitting distance of the Liberals in Ontario and Atlantic Canada.

The degree to which they trail Trudeau's Liberals right now is much larger than it was when they trailed Thomas Mulcair, Michael Ignatieff, and Stéphane Dion in the past. To find the last time the Conservatives trailed anyone by 12 points in national polls, you have to go back to 2005 when the party was sitting on the opposition benches. The numbers can be waved off as being two years from the next election, but it would take willful blindness not to recognize that the Conservatives are currently in the deepest hole they have found themselves in since coming to power more than seven years ago.

23 comments:

  1. Once again Eric, I really appreciate the intelligent analysis. You break down the data, compare apples to apples, and put the uncertainties and limitations of the data in perspective. I hope every political reporter reads your blog.

    I sometimes despair when I see the pundits superficially comment on the results of individual polls, without discussing their usefulness and limitations. They are journalists and are suppose to be able to communicate this succinctly to the public, and fail to do so everytime.

    If I could add some commentary, this looks like a perfect storm for Trudeau. He does seem to be taking on a Teflon coating. It would be interesting if there were a shift from the undecided column to the Liberals, as it would indicate that Trudeau is engaging the electorate to participate in elections. It looks like the Green protest vote, particularly in Ontario, is shifting Liberal.

    That being said, the numbers here shows the resilience of the NDP in Quebec (they are a solid second, with the Bloc and Tories fading) and appear to be the alternative to Trudeau's Liberal pedigree at the moment.

    The Conservatives show resilience in Ontario. They appear to have a solid baseline of support here at 30%, while the other parties seem to show volatile support.

    ReplyDelete
  2. Boy we live in interesting times, politically. I don't think Liberal supporters can necessarily take these numbers to the bank, Trudeau honeymoon etc., however the days of the LPC stuck at 19 per cent on a good day are over. The Libs are also being helped by the general and growing distaste for the Harper Conservatives. The only thing that might save the CPC is if they replace Harper with a politically savvy, hard-nosed campaigner. Take for example Today's BC Liberals. Clark saved their bacon, and someone else might do the same for the CPC if they can convince the public they are the New, or Today's, CPC. And of course, keep pouring out the Trudeau attack ads.

    ReplyDelete
    Replies
    1. Strong economy. Strong tomorrow.

      Delete
    2. Great slogan, Ryan. I like how you right-wingers think. Trudeau should make sure to use it, him being a right-winger and all.

      Delete
  3. Eric,

    Do you use EKOS' likely voter numbers or their unweighted samples in your average? I realize they don't give a good breakdown for the likely voter numbers, but presumably one can generate one from the unweighted sample with some basic assumptions. The reason I ask is because I have grave doubts about the accuracy of EKOS' unweighted numbers. As I'm sure many here do.

    ReplyDelete
    Replies
    1. I use the samples for the general population (these are not 'unweighted', they are meant to be a reflection of the entire population).

      Delete
    2. Fair enough on the terminology. You get my point though. Even EKOS is upfront that their general population poll is not reflective of how people would vote. Their likely voter model is.

      Delete
    3. True, but outside of elections I think it is better to look at the general population.

      Delete
    4. Fair enough. On the seat projection side you weight down the third party numbers anyways don't you?

      Delete
    5. No, what do you mean?

      Delete
    6. I thought your model accounts for the fact that the third party tends to under perform the polls, and that the first party tends to overperform?

      Delete
    7. Not when I'm looking at polls - then I just translate the numbers into seats (i.e. if the poll was the election result).

      When I make 'official' site forecasts for upcoming elections, then yes I do make those kinds of adjustments.

      For this post, for example, I'm reporting only on what the polls are saying and turning that into seats, not making estimates on how the polls might be wrong.

      Delete
  4. Are these calculations based on the current Parliament or the new Parliament (the ridings of which are being redistributed)? If it is the latter, could I ask where the data for the redistributed seats is located?

    ReplyDelete
  5. looks like Atlantic Canadians have really settled on their guy. Those numbers are astronomical.

    The Liberal surge is making all your line graphs look funny. It doesn't look smooth or "natural" if you know what I mean.

    ReplyDelete
  6. Eric,

    I know you make reference to the fact the NDP seat projection in Quebec is open to change depending of the specific variables of a general election and the particularities of ridings in that province.

    Not to point fingers or question the accuracy of your seat projection but, I think I should report that when using your Quebec numbers on an academic seat projection algorithm (which is purely a swing model) the Liberals captured every seat in Quebec.

    I recognise a difference in methodology is likely the cause of this divergence, however, I thought the divergence exceptional since, usually when inputting your numbers the difference in seats is small often +/- 5 seats. In the latest example the deviation was 43 in Quebec alone.

    ReplyDelete
    Replies
    1. Not knowing what model you are using, I cannot comment. I don't know how that would be possible, though.

      Take the riding of Abitibi-Temiscamingue. With my model, I get 42% for the NDP, 27% for the BQ, and 20% for the LPC. The Liberals have only grown from 14% to 41% in Quebec, growth of less than three times. Multiply the 5.9% the LPC had in that riding by three, and you only get 17.7%. How would it be possible to award the Liberals this riding?

      Even with a uniform swing model it is not possible. The Liberals increased 26.8 points in Quebec, so that puts them up to 32.7% in this particular riding. The NDP has dropped 15.2 points, which would drop them to 36%. They still win it.

      Are you sure you did not make a mistake?

      Delete
    2. A drop of 15.2 points would put the NDP at 25% actually, handing it to the Liberals.

      Delete
    3. No, the NDP had 51% in that riding.

      Delete
    4. Hi Eric,

      Thank you for your reply.

      I carried out my experiment again using your Quebec numbers and the result was the same. In Abitibi the Liberals would receive 35.4%; NDP 33.1%; BQ 23.6; Conservative 6.5%. As I said before my understanding is this is a pure universal swing model.

      In any case I thought it interesting that two seat projections (using different methodolgies) would vary so widely.

      Cheers,

      Bede

      Delete
  7. Your projection model seems to be forecasting something like a repeat of the 2000 election in Quebec, with the NDP in place of the Bloc. Like 2000, the Liberals seem to be currently somewhere in the low 40s. What's different this time is that the Bloc was also at 40 while the NDP are polling in the low 30s or even high 20s. Moreover, the NDP are likely to waste some of those votes in heavily Liberal anglophone areas while the Bloc vote was concentrated in winnable francophone seats. It seems improbable to me that, given these differences, we'd see a seat result similar to 2000. I'd be fascinated to hear your views on these similarities and differences.

    ReplyDelete
  8. Hi, I come from Australia and I am doing a study into Canadian elections and wished to know how common is vote splitting where a notionally left seat for example, is lost to the right because of a split vote in first past the post?

    ReplyDelete
    Replies
    1. In Canada the Liberal party is only nominally "left". Throughout most of their history, at least since Mackenzie King, they have been the party of Bay St (ie. the Toronto business elite). In this sense our two major parties are a Conservative party and a small "c" conservative party. Generally those two parties (with the last federal election being a notable exception) win two thirds or more of seats in the House of Commons. The Bloc while adhering to "left wing" social policy is a conservative-nationalist party who seek to create a nation-state for the Quebecois people.

      Historically, Canada does not follow the traditional left-right spectrum instead political cleavages have traditionally been along linguistic, religious, ethnographic lines or continental integration v. Canadian nationalism.


      Delete

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.