Wednesday, August 22, 2012

Charest's Liberals in the lead?

As I work to include the latest poll by Forum Research for The National Post into the projection, I invite you to read my take on the surprising results in my article for The Globe and Mail.

The reaction to the poll has been pretty skeptical in the Twittersphere. I have to admit that such a dramatic and unexpected turn of events leads me to raise an eyebrow as well, but we should wait and see what the other active firms in Quebec have to say before passing complete judgement on the poll. Then again, as I mention in the article, the three one-on-one debates could change things on their own and thereby prevent us from knowing whether Forum was capturing a real surge in support for the Liberals or just a statistical wobble.

Practically speaking, the projection will swing pretty wildly due to this poll. The most recent survey before Forum's is older by five days (in terms of the median date used for weighting), meaning Forum's poll will take up a lot of space in the projection. It will be more interesting to see where everything stands after CROP and Léger have reported, as it will give us a chance to 'triangulate' things, but if Forum is indeed capturing a shifting tide then the projection will have swung in the right direction. We shall see.

35 comments:

  1. The PQ losing a few points to QS would not be "shocking" per se, but a Liberal surge?

    This is surprising. A statistical aberration maybe? Or could it just be that many Liberal supporters came back from summer vacation?

    ReplyDelete
    Replies
    1. Ha! Possibly. Or that PQ supporters left for their vacations. This being the summer, I imagine the people who are home on any given day changes quite a bit more than during the fall or winter.

      Delete
  2. Forum Research? c'est la firme qui tente d'avoir mes opinions avec un message informatisé est en ANGLAIS et auquel je refuse de répondre par principe?

    ReplyDelete
    Replies
    1. Je trouve cela difficile à croire........

      Delete
    2. Néanmoins, Forum avait jusqu'ici trouvé le PQ en tête dans tous ses sondages depuis le début de la campagne, avec plus d'appui même qu'ont trouvé CROP et Léger (34-39% vs. 32-34%). Allez voir:

      http://en.wikipedia.org/wiki/Quebec_general_election,_2012

      Comme toujours, il va simplement falloir attendre de voir si les autres sondeurs confirment la tendance.

      Dom

      Delete
  3. Or maybe Eric it's just that people are beginning to realise that the PQ alternative has essentially NO experience in governing ??

    And lots of "foot in mouth" experience ??

    ReplyDelete
  4. I work in a polling firm currently polling the Quebec election and I can tell you, even though I don't have results, that systematically the people who say they will vote for Legault are also the ones saying they are not firm on their decision yet. Some admit they are waiting to see the debates. Those who prefer Charest or Marois are firm. That could indicate that the people who are flirting with the idea of voting for Legault will be likely to swing back to Charest come September 4th provided they don't appreciate Legault's performance in the current face offs...

    ReplyDelete
  5. I'm no fan of the PQ, but its very wrong to say they have "no experience in government". The PQ was the government of Quebec from 1976 to 1995 and again from 1994 to 2003...Marois herself has been minister of almost everything and there are lots of other ex-cabinet ministers running for the PQ.

    ReplyDelete
    Replies
    1. DL

      Right off the bat your numbers just don't cut it nor do they add up !!!

      Secondly Charest is finishing his 3rd FULL term !! Or is it 4th??

      So that means the latest PQ would be 2000 and again your numbers don't work !!

      Delete
    2. DL meant that the PQ was the government from 1976 to 1985, (not 1995).

      Charest is not finishing his third FULL term, as the second term (from 2007 to 2008) was short due to a minority legislature.

      The Liberals have won the last three elections, in 2003, 2007, and 2008. The PQ was last in government in 2003.

      Delete
    3. So he garbled it !!

      No excuse

      Delete
    4. I'd say that a typo is indeed a good excuse.

      Delete
    5. Peter, the point was, you claimed the PQ had no experience in government. DL effectively refuted that.

      Delete
    6. Come on Peter, everyone makes mistakes, let's show some tolerance and respect here.

      Delete
  6. Would be interesting to know who they polled. There's no information on the method they used, like were they mostly anglophones?

    It is kind of random to publish the final result without providing any other useful information to judge the quality of their poll.

    ReplyDelete
    Replies
    1. The information is on Forum's website, it has the usual proportions of anglophones and francophones that you would expect.

      Delete
  7. Eric,

    Can it be right that in your table the poll average for CAQ (26.3%) is higher than the Popular Vote Projection High (25.6%)? I can't immediately think of a good reason for why that would be.

    Geoff

    ReplyDelete
    Replies
    1. It's the math. The CAQ is adjusted downwards (along with QS and ON), and the projection ranges are based on the volatility in the polls. Since the polls were not very volatile last week, the range is not very different from the projection.

      The poll average is there for information purposes only.

      Delete
    2. OK, thanks, I think I see, though it seems odd that the average of recent polls could be higher than the projected best case for a given party. Presumably the key is that the period over which the poll average is calculated is shorter than the one used for the projections.
      Geoff

      Delete
    3. No, it is the same. The model believes that the polls will over-estimate the CAQ, even in a base-case-scenario (at least with this level of volatility).

      Delete
    4. Thanks again. What accounts for the model's scepticism about CAQ support in the polls? Is it marked down only for variability, or is there also a discount for being unproven in actual elections?

      Delete
  8. Translation from Anon above:
    "Forum Research? Is that the firm that tries to get my opinion with a computer questionnaire in English which I refuse to answer on principle?"

    If this is indeed their methodology, no wonder the results seem odd...

    ReplyDelete
    Replies
    1. Forum does not ask their questions in English only.

      Delete
  9. Il y a un facteur à ne pas oublier. Il est de bon ton au Québec, du moins parmi les Québécois francophones, de démoniser le PLQ, à tort ou à raison, et par extension, ceux qui votent pour eux. Ma page Facebook est remplie d'affichages d'amis qui insultent les "idiots" qui considèrent voter pour le PLQ, ce parti "corrompu qui trahit le Québec" etc. Je suis persuadé que beaucoup de voteurs du PLQ disent voter pour un autre parti lors de sondages avec un intervieweur humain, mais se laissent aller quand c'est une machine. Le véritable score du PLQ est sans doute à mi-chemin entre les sondages classiques et ceux de Forum. On va avoir des surprises le 4 septembre.

    ReplyDelete
    Replies
    1. En effet, c'est très possible qu'à cause de l'hystérie anti-Charest, certains électeurs soient discrets et ne donnent pas leur vraie opinion. Dans ce cas de figure une majorité des "indécis" voterait pour le PLQ.

      Delete
  10. Et si marois avait fait peur avec son référendum et sa nationalité québécoise?

    ReplyDelete
  11. I believe it's a combination of undecided and soft swing voters howevering back n forth between he CAQ and Liberals.

    Tired Liberal voters are the ones thinking of voting for the CAQ but they are also voters who are ready to revert back to voting Liberal to block a PQ majority.

    The undecided have remained relatively large as and they are the ones deciding this... not the base

    ReplyDelete
    Replies
    1. exactly what i will do.

      Delete
  12. All experts tends to say that this poll is necessarly wrong and to wait to see the results for Crop and Leger this weekend.

    ReplyDelete
  13. Commentaire d Alain Castonguay


    « Je le dis depuis plusieurs semaines et lors de chaque sondage de la firme Research publié dans le National Post. Ce sondeur a souvent des chiffres complètement différents des autres et n'a pas de bureau au Québec. Il a de la difficulté avec la pondération des francophones et il s'agit de sa première élection ici. Il avait eu des résultats désastreux en Alberta, estimant que le Wildrose Alliance allait remporter les élections avec une majorité écrasante.

    ReplyDelete
    Replies
    1. Au contraire, Forum a été le *seul* sondeur lors de l'élection en Alberta à détecter le rapprochement de dernière minute entre les PC et le Wildrose. Allez voir:

      http://en.wikipedia.org/wiki/Alberta_general_election,_2012#Opinion_polls

      Dom

      Delete
  14. Francoise David is more likable and straightforward than Pauline Marois. The Quebec Solidaire platform on soveriegnty is also more clear than that of the PQ. It's not surprising seeing younger progressives deserting the PQ after such a strong performance by Quebec Solidaire.

    I'd be interested on how the different polls track the undecided. Perhaps this represents a falling of PQ voters into the undecided column, something that would boost the Liberal proportion among decided voters.

    ReplyDelete
  15. Geoff in Suroit22 August, 2012 16:41

    I think this may be a tide turning poll. I believe the debate where Charest pulled out the 2006 report, may have had something to do with it. Most people support the tuition increases, believe that the PLQ is the best to run the economy. Corruption has been in Quebec since Duplessis and both the PQ and PLQ have had their moments. Personally there isn't much policy that the PLQ implemented that I consider bad. Quebec does not want Marois, and I think they want to see Legault before passing judgement

    ReplyDelete
  16. I really find it strange that your projection is so overly sensitive to a single poll. Why would it make such a huge swing in the projection based on one poll by one company? It seems disappointingly easy for the projection to be fooled by a single outlier poll, something I don't think should be given that much stock given how unlikely the result seems.

    ReplyDelete
    Replies
    1. During the federal election, when there were multiple polls out every day, this was never a problem. If the firms active in Quebec reported in a more staggered fashion, it would also be less of a problem.

      Instead, we have had Forum on Wednesdays and CROP/Leger on Thursday/Friday. Meaning that when Forum reports on Wednesday, there has been a big gap since the last polls and so it takes a up a lot of room in the projection - until CROP and Leger report, at which point the projection is more balanced.

      The projection ranges are one way of keeping this in check. But let's imagine that the election was tomorrow - would it be right to assign relative equal weight to the Leger and Crop polls taken last week to this new set of information? Both firms were in the field more than a week ago.

      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.