Thursday, November 7, 2013

Liberals making gains in by-election polls

With all the brouhaha in Toronto and the Senate, perhaps we need to be reminded that an election is actually on-going in the four federal ridings heading to the polls on Nov. 25. And yesterday evening, Forum Research came out with their latest set of numbers for those four by-elections, showing the Liberals leading in three of them and making (mostly modest) gains across the board.
Forum was last in the field for these by-elections on Oct. 17-18, just before the elections were officially called.

The Liberals made their biggest gain in Bourassa, where Emmanuel Dubourg picked up nine points to reach 56% support. That jump was outside the margin of error, and apparently came almost entirely from the Greens. With Georges Laraque off the ballot and Danny Polifroni in his place, support for the Greens fell 10 points to just 2%. I suppose that demonstrates just how much Laraque was worth, as his party was polling at 3% in the riding earlier this year before he threw his hat (temporarily) in the ring. It has to be disappointing for the Greens, who have put a lot of resources into the riding.

Support for the NDP's Stéphane Moraille increased by a single point to 19%, while Daniel Duranleau of the Bloc Québécois was up two points to 17%. That also has to be disappointing for the New Democrats, who had hoped to be able to wrest the riding away from the Liberals, or at least keep it competitive. Instead, Dubourg led by a wide margin among most demographics and the party has actually been increasing its support over the last two polls.

The Liberals also led in Brandon-Souris, up a single point to 40%. The Tories' Larry Maguire was unchanged at 35%, while Cory Sczepanski was down two points to 10%. Forum did increase its sample size in this riding from last time, and still found a very similar result to the mid-October poll. Rolf Dinsdale led among most demographics as well, suggesting his support is wide as well as deep.

In Provencher, Ted Falk of the Conservatives dropped three points to 53% but continued to hold a lead over Terry Hayward of the Liberals, who was up five points to 34%. Falk enjoys a good lead among all demographics. Natalie Courcelles-Beaudry of the NDP was down a single point to 8% in the riding.

And in Toronto Centre, the Liberals' Chrystia Freeland was steady with a gain of a point to 46%. Linda McQuaig picked up five points - outside the margin of error - to close the gap with 35% support. Geoff Pollack of the Conservatives was down three points to 15%. Significantly, Freeland had a big lead among older residents of the riding (i.e. voters), while McQuaig did best among those with a household income of $40,000/year or less (not exactly middle class). But McQuaig is steadily making gains, as the candidate-less party had 20% support earlier in the year. That increased to 30% in October, and is now at 35%. The Conservatives seem to have been dropping as a result (indirect or not).

Unfortunately, Forum did not release a breakdown by past federal vote, so it is impossible to check the weightings as I did in my last report on Forum's by-election polling.

Below are the weighted averages of the polls in these ridings. Forum's are the only ones in the game, and the 67% rule has been applied to the weightings to ensure that no single poll takes up the entire calculation. As previously explained, the 67% and 95% confidence intervals take into account the past degree of accuracy of by-election polls.

Dubourg is considered virtually unbeatable in Bourassa according to Forum's polls, as he still wins handily even at the lower end of the 95% confidence interval.

Brandon-Souris remains a toss-up leaning Liberal, as both confidence intervals put Maguire and Dinsdale in a position to come out on top.

Falk should have no problem holding the riding, but at the 95% confidence interval Hayward is getting perilously close to being in range of a surprise upset.

Freeland's lead remains wide enough that she should be able to win, but McQuaig is close enough as well that she is more than in the running to take it on Nov. 25.

It will be interesting to see where the numbers go from here, as minds focus a little more as voting day approaches. In the end, these were all relatively comfortable ridings for the incumbents so it should come as little surprise that the Liberals are favoured in two of them and the Conservatives in Provencher. But Brandon-Souris is definitely the race to watch - and Toronto Centre is still potentially up for grabs.

25 comments:

  1. While not exactly a "given" these results seem to be roughly were expected ?

    ReplyDelete
  2. Fascinating. The close race in Brandon reminds me of how Calgary Centre very nearly fell to the Liberals a year ago, and probably would have had the Greens not come out of nowhere to tally up 26% of the vote themselves!

    If the Liberals continue to nominate strong candidates across the prairies (including Alberta), 2015 may turn out to be even more interesting than it already looks.

    ReplyDelete
    Replies
    1. Turnout is a bigger factor in the by-election than the general election. It's quite possible that without a Green on the ballot many of the people who voted Green in Calgary Centre would have just stayed home.

      Delete
    2. I agree with you there, but of course not all 26% of them would have. Harvey Locke, the Liberal candidate, only needed a few more points to win, and the irony of the situation is that Locke was arguably a bigger environmentalist than the Green Party candidate, and certainly one of the country's most prominent. His "green" credentials are indisputable.

      Yet I think the perception at the time that it was a three-way horse race and that the Green candidate also had a good shot at winning prevented any serious strategic voting.

      Delete
    3. I've been living in Calgary for the past few years, and it irks me to no end that people still have this perception that the Green candidate drew votes away from the Liberal in Calgary Centre. That's not how it happened, The Green candidate was very credible, and he got most of his support from people that previously voted Conservative but were growing dissatisfied with the government. They wanted to protest vote but they sure as hell weren't going to vote Liberal.

      The Greens took votes away from the Conservatives and that's why the Liberals managed to get remotely close in Calgary Centre.

      Delete
    4. That is your view but also a minority view. Liberals have the ability to elect MPs in Calgary and will do so next election.

      Delete
    5. I agree with Matthew. I lived in Calgary at the time and much of the Green and Liberal vote were from dissatisfied Conservative voters. I do not think this the minority view, at least not with people in the know.

      I am sick and tired of primarily Liberals and NDPers denigrating Green votes. Many Green voters vote Green due to is the past performance of Liberals and Dippers or their current policies: The Liberals are a small "c" conservative party with strong links to industry and Bay St. whereas; the NDP favour union jobs in resource industries over the environment. Neither party has a particularly strong environmental record or appeal to the environmentally driven voter.

      Delete
    6. Agreed for the most part there Bede. Voters aren't as partisan as people potray them. Very few people actually think or say "I'm a Conservative" or "I'm an NDPer." People vote for who they like, and they may like a given candidate for complex reasons.

      Delete
  3. This comment has been removed by the author.

    ReplyDelete
  4. The Notion of the Liberal Party of Canada being only a few points more popular in Toronto Centre than Brandon-Souris is a strange and unnerving new world

    ReplyDelete
    Replies
    1. I'd imagine they're the second choice of many more in Toronto Centre than Brandon-Souris.

      Delete
  5. Does anybody know what the record of seats staying in a party in bye-elections is ??

    Eric ??

    ReplyDelete
  6. This *just* came out - A very interesting article from CEO David Coletto of Abacus Polling saying that the Forum polls for Toronto Centre have been drastically underpolling anyone under 45, and that using the data balanced against real population it would be a viritual tie (either slight NPD or Liberal lead at this stage)

    "Clearly, it’s sample is heavily biased towards older voters. Statistical weighting can help solve this problem."

    He comes up with NDP = 39.7% vs Liberal = 39.3% based on the real census numbers, or Liberal = 42% vs NDP = 38% if looking at 2011 vote numbers.

    Reference: http://abacusinsider.com/politics-public-affairs/battleground-toronto-ce...

    ReplyDelete
    Replies
    1. Yes, it's a good analysis (I re-tweeted it on Twitter). I've been banging on about how Forum over-samples older people for some time, but I imagine they have a turnout model in place because it is hard to work out how they weigh their polls.

      In the end, not including cell phones and under-weighing young people might get you closer to the result. But that doesn't say anything about the quality of the poll, though.

      Delete
    2. Elections Canada indicates slightly over 40% of 18-34 year cohort vote, whereas upper cohorts turnout in greater numbers. For example,The 35-44 year cohort has a turnout rate of almost 55% with turnout increasing until the 75+cohort.

      Given that by-election turnout is often half of general elections at 30%+/- and the NDP in particular receives a disproportionate share of its vote from the cohort least likely to vote (the BC election a large number of people who indicated to pollsters they would vote NDP did not turn out). I think some form of scaling should occur or perhaps weighted based on a respondents probability they will vote.




      Delete
    3. If a pollster is using a likely voter model, they should state so.

      And generally, I think an approach where the poll itself tries to determine likelihood of voting is much better than a model based on assumptions from past elections.

      Delete
  7. FYI, we know that Tory support tends to skew very old and so in a contest that was CPC vs NDP or CPC vs Lib - you might expect the Tories to benefit from the turnout factor...but in a contest that is purely NDP vs Liberal - its less clear how the age skew affects things since both the NDP and Liberals tend to have support that skews young compared to the Tories...it should be pointed out that in the byelection this summer in Ontario, the NDP overperformed and the PCs underperformed compared to the final polls in most of the seats - so go figure.

    ReplyDelete
    Replies
    1. Agreed. Voter enthusiasm (for positive or negative reasons) is a key factor. When Bob Rae's NDP won in Ontario, they obviously overcame this problem - precisely due to the enthusiasm to oust Peterson (and, I would argue, because people were eager to try something different from the two 'old parties'...). [What Rae did with this victory is another issue...]

      Delete
    2. NDP support skews very young whereas Liberal voters are more evenly distributed among the population although generally speaking the probability of voting Liberal increases with age. Among the >35 turnout is about 40% whereas in the 45-54 cohort it rises to 55% and in the 65+ cohort 75% according to Elections Canada. So in a NDP-Liberal contest all things being equal the Grits have the advantage in terms of turnout and hence vote.

      Ekos often provides voting intentions by age. For their latest poll the >35 yrs quintile was L: 34%, C: 18%, N:34%; whereas the 45-54 yrs. quintile was; L:39%, C: 28%, N:19%.

      Delete
  8. Hi Éric, just a heads-up that there some ON provincial voting intention results from Forum concealed in the following article:

    http://www.thestar.com/news/queenspark/2013/10/29/better_canada_pension_plan_more_popular_than_proposed_ontario_version_poll_suggests.html

    Dom

    ReplyDelete
    Replies
    1. Thanks, missed that. Forum does not have the details up on their site, will have to get them direct.

      Delete
  9. Hi Éric, two suggestions for the By-election barometer:
    1) List the date of the by-election if it's been called, or clearly say "not yet called" if that's the case. I love being able to use your site as a go-to for "What to watch next", and that would be really helpful for me - maybe even in your summary table for the riding?
    2) When there are no riding-specific polls yet, that square still ends up with a colour. Should it maybe be blank, or grey?

    ReplyDelete
    Replies
    1. I do outline the dates of the upcoming by-elections in the paragraph before I get into the forecasts. ("Upcoming By-Elections")

      For your second point, I just put the colour of the incumbent. I'll see what it looks like without putting some colour when I next update.

      Delete
    2. Okay, I see what you're saying - Thanks for pointing me to that. I still think that there might be a way to integrate it into the summary table of polling results, but that's me being an armchair quarterback - your site is great, and effective as is, regardless.

      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.