Friday, August 31, 2012

PQ on verge of losing majority

Polls from Forum Research for The National Post and CROP for La Presse have been added to the projection. Together, they suggest that the Parti Québécois's quest for a majority is on a knife's edge, and that the role of the Official Opposition is still very much up for grabs.

The projection awards the Parti Québécois 33.7% support, compared to 28.1% for the Liberals and 26% for the Coalition Avenir Québec. In fourth is Québec Solidaire, at their highest point in the campaign with 7.8% support.

This would deliver 64 seats to the PQ, 33 to the Liberals, 26 to the CAQ, and two to Québec Solidaire. As 63 seats are needed for a majority government, the PQ is really on the edge. If the projection is off by two seats, which would be a great result, it would mean the difference between minority and majority.

The ranges show, however, that the PQ is the only party in the running for government. They could win between 54 and 73 seats, straddling the majority line almost exactly. The Liberals, with between 27 and 43 seats, are well placed to form the Official Opposition (and they can thank Montreal for that), but the CAQ could also take on the job with between 19 and 32 seats.

Québec Solidaire, with between one and two seats, and Option Nationale, with potentially one seat, could hold the balance of power.

The CAQ has been making gains in central Quebec and the suburbs around Montreal, putting a lot of pressure on the Parti Québécois. They are doing very well in Quebec City but are not strong enough in other parts of the province to have a shot at forming government. The Liberals, with their solid base of support in Montreal and the Outaouais (22 of their 33 seats), have a head-start on the CAQ.

The turnout adjustment benefits the PQ and the Liberals only slightly in terms of seats. Without the adjustment, the poll averages alone would still deliver 32 seats to the Liberals and 61 to the Parti Québécois. That means a minority, but with the two seats from Québec Solidaire the PQ could make it work. The CAQ, with 30 seats, still ends up in third place even with a second place finish in the popular vote.
The unexpected CROP poll of this morning shows very little change from their last poll of Aug. 24-26. The PQ is down one point to 32% while the CAQ and PLQ are steady at 28% and 26%, respectively. With a two point gain, Québec Solidaire is now at 9%.

All of these shifts are within the margin of error, however, suggesting that things are now solidifying. The CAQ needs to continue its momentum in order to have a shot, but that does not seem to be in the cards for them.

But it is worth noting that while the shifts in support are statistically insignificant, the lead the PQ holds over the CAQ is as well (just). We also can't say with absolute certainty that the CAQ is ahead of the Liberals (though we can say that the PQ is). It makes for a potentially interesting finish on Tuesday night.
Forum fell into line with everyone else after their stand-out poll from last week, but the firm still seems to be showing better numbers for the Liberals. Perhaps the IVR method overcomes a "shy Tory" factor for the PLQ. We will find out on election night.

Since that last poll of Aug. 20, the PQ has gained four points and leads with 33%, followed by the Liberals at 28% (-7), the CAQ at 27% (+3), and Québec Solidaire at 8% (-1).

Together, these two polls tell a similar story of a close three-way race. The PQ has the advantage but neither the CAQ nor the PLQ are taking on runner-up status. The CAQ is ahead in Quebec City and the PQ elsewhere, while QS is doing better than they were before the debates.

We will have to wait and see what Léger Marketing says with their final poll, expected this weekend.

Riding polls

A large amount of riding polls were released in the last few days, and all have been added to the projection. Let's go through them quickly.

A poll for Le Soleil by Segma in Taschereau found Agnès Maltais of the PQ well ahead of Clément Gignac of the Liberals, with 41% to 19%. QS and the CAQ tied for third with 16% apiece. The projection also showed Maltais as the likely winner.

Another Segma poll for Le Voix de l'Est in Brome-Missisquoi found a close three-way race. Benoit Legault of the CAQ came out ahead with 32% to Richard Leclerc's 31%. The incumbent Liberal, Pierre Paradis, was in third with 27%. The projection also shows a close three-way race, but gives Paradis the advantage.

A poll in Papineau by Segma for Le Droit put Jean-François Primeau of the PQ ahead with 36%, followed at length by Alexandre Iraca of the Liberals with 27%. It confirmed the projection's prediction that Papineau would go PQ, but it did not give Primeau such a large lead in the formally safe Liberal riding.

Three polls by Forum Research for the Stantead Journal showed close races in Estrie: 36% for the Liberals to 35% for the PQ in Orford, 39% for the PQ to 37% for the Liberals in Saint-François, and 41% for Serge Cardin of the PQ to 39% for Jean Charest in Sherbrooke. That last one is of some interest, considering the leads that Segma has given Cardin in two polls.

Moving on to the Gaspésie, two polls by Segma agreed with 308 that the PQ could win Bonaventure and Gaspé, but the poll for Graffici put the Liberals much further behind in the latter: 57% for Gaétan Lelièvre in Gaspé to 20% for the incumbent Georges Mamelonet. It showed a closer race in Bonaventure, with 41% for Sylvain Roy of the PQ to 34% for Damien Arsenault of the PLQ.

Another poll in Saint-François, this time by Segma for La Tribune, gave the PQ's Réjean Hébert a 20 point lead over Nathalie Goguen, with 46% to 26%. That is much more in line with what the projection had.

And finally, five polls this morning by Segma for Le Quotidien in Dubuc, Chicoutimi, Jonquière, Roberval, and Lac-Saint-Jean confirmed that the PQ is poised to sweep the Saguenay-Lac-Saint-Jean. Interestingly, however, in all of the ridings except Roberval the CAQ has moved into second, ahead of the Liberals. On average, the CAQ gained six points in these ridings since Segma's last poll of the region between Aug. 11-16. Both the PQ and the Liberals have lost three points. This is a strong indication of how the Liberals are suffering in francophone Quebec, but also how the PQ is relatively stagnant. Remember that the PQ is likely to win the election on Tuesday with less of the vote than they took in their 2008 defeat.

26 comments:

  1. Éric, out of curiosity, have you considered the possible ramifications of the record turnout in advance polls (15.6%) on the ability to accurately predict the outcome of the election, in particular should the polls show any noteworthy shifts in support between now and Sept. 4?

    Dom

    ReplyDelete
  2. It looks like Quebec City is swinging back to the PLQ.

    ReplyDelete
  3. Is that a coalition of Socialist Separatists and Separatist Socialists I spy?

    ReplyDelete
    Replies
    1. Sorry but PQ barely rank as social democratic, let alone socialist, these days. But your scenario sounds delicious.

      Delete
  4. Obvious the Quebec election takes priority, but any chance you'll look at the two recent Ontario polls by Nanos and Forum? The Nanos one is especially interesting given the low result for the NDP and the fact that the Liberals and the PCs are in a virtual tie.

    Also, any chance the by-election barometer will be updated soon? It'd be interesting to see if Kitchener-Waterloo is still a strong PC seat, and in terms of federal seats Victoria would be interesting to see as well.

    ReplyDelete
    Replies
    1. I wouldn't put much faith in Nanos' numbers. They seem to be in the realm of outliers and almost consistently seems to rank the Liberals much higher than other firms do.

      Not that they are wrong or their numbers incorrect, but I think more polls would have to come out to verify the numbers before I can take his poll seriously.

      Delete
    2. It's frustrating isn't it? Nanos has a great track record, yet they are disagreeing so much with pollsters with equally good track records. Someone is wrong, but we won't know who until 2015.

      Delete
    3. And Eric, when all the smoke clears from the Quebec election, perhaps keep an eye open for a BC poll from AR which should be surfacing in the next few days. It was an on-line poll of registered respondents, but asked some interesting questions beyond the usual who would you vote for. For example they asked about perception of the Gateway Pipeline project, which is a huge political hot potatoe out here. And they also asked voting preference using scenarios where the BC Liberal party had several different leaders including Stockwell Day. Should be a hoot.

      Delete
  5. It seems bizarre that one poll shows the Liberals at 26% support with 19% among Francophones and 54% among Non-Francophones and yet another poll has them just two points higher overall (28%) despite much bigger numbers in both of those demographics (26% and 67%). If those demographic numbers are correct, shouldn't the Liberals be closer to something like 34% overall in the Forum poll?

    ReplyDelete
    Replies
    1. That's a very good point. Are they not re-weighting the numbers based on linguistic demographics?

      Delete
    2. And you are right, number are a bit off... My estimation are that, estimation, but just considering English voter, and thus underestimating the advantage of PLQ in the other (you have people not speaking mainly French or English and they are starting to be more that just 1 or 2% of the vote), adjusted number (for Forum) should be around 32.2% for PQ, 29.7% for PLQ and 27.2% for CAQ.

      Of course this is not scientific. But if that true, then even Forum underestimate the PLQ (and by a solid 2%). And by experience, PLQ is always down of 3% in polls, their vote is more solid that the others. On the other hand, the PQ often manage to do up to 3% less that what they were predicted... Non-scientific, but that would point to a PQ-PLQ tie and not the PQ being so much in advance.

      Definitely, or I'm completely off for my statistic, but then newspapers are a big bunch of liar (big surprise), or polls have some internal adjustment (we don't know how they distribute uncertain voter, maybe they do it proportionally (PQ = 30% so they got 30% of uncertain for an example, or they don't.).

      Also it's interesting to note that around the same time in the 2007 campaign, the ADQ was about where the CAQ is now in polls, and they got official opposition, missing power by about 6 seats (41 to 47 if I remember). I don't know for the 450 region, what the polls are telling, but people like Jean Lapierre, who got there in person said that in majority people are going to vote for the CAQ, not unlike again the ADQ in 2007.

      (approx. pop of 750k Québécois with English as their mother tongue, if they vote in the same proportion that everyone and you have 6 millions of voter (15% ~ 900k (in advance polls) then they are around 560k of English voter for a weight of 9% globally, by extension francophone are 91%)

      Delete
    3. It's just a very strange mix. Assuming every voter is either Francophone or Non-Francophone in each poll (and maybe this is a bad assumption on my part), to make the CROP numbers work you need a split of 80% "F" and 20% "NF", which seems reasonable to me. In order to make the Forum numbers work you need a split of 94% "F" and 6% "NF", which strikes me as being far too low unless they count everyone who is fluent in French as Francophone. Either way, it seems that the two polls either managed to get a very different mix or are using quite different definitions. Do you know which is the case, Eric?

      Delete
  6. Eric, I've wondered for awhile - how do you add the riding polls to your projections? (Sorry if I missed this somewhere else on the site)

    ReplyDelete
    Replies
    1. They are not weighted for more than half of the projection in any riding, and are weighted less if they have a sample of under 500.

      They are used as a new jumping-off point for the projection in each riding (at least for the amount of weight it takes up). When the poll is added, it is added without any adjustment. But as the projection in the region shifts, the support assigned to each party in the riding poll will shift accordingly.

      So, for example, let's assume a poll puts the PLQ at 20% in a riding when the PLQ is at 25% in that region. If the PLQ then drops to 15% in the region (a loss of 40%), the PLQ's result in the poll will be reduced by 40% as well (in this case, to 12%).

      Delete
  7. Keep in mind that the last time there was a similar 3-way contest in a Quebec election in 2007 - the final CROP and Leger polls each gave the ADQ 26% - and on election night the ADQ got 31% - while the final CROP and Leger polls gave the Quebec Liberals 35% and they ended up with 33% - so it is a bit of a myth that polls in Quebec always underestimate the Liberals.

    ReplyDelete
  8. Thanks for all this analysis! Quick question: why do you claim that the PLQ has Montréal to thank for its shot at forming the official opposition, when they have around the same score (26%) in Montréal RMR, Québec RMR, and in the Rest of Québec?

    ReplyDelete
    Replies
    1. It's not the Montreal RMR, its the Island of Montreal. The Liberals win most of their seats there.

      Delete
  9. For a party that could have won a comfortable majority, the PQ ran a very weak campaign. They focused too much on Quebec nationalism and less on other issues, particularly the economy and corruption.

    They may still manage to win a slim majority. But a majority government with only 33-35% of the vote is a very weak mandate to implement any policies regarding the Quebec identity. A PQ majority government (or minority propped up by the QS) would result in 4 years of dithering in Quebec.

    ReplyDelete
  10. Bonjour,

    Pourquoi QS, le PQ et le PLQ étaient à égalité dans la marge d’erreur ce matin? Cette projection était déjà à jour avec le dernier CROP, avez-vous changé des coefficients?

    ReplyDelete
    Replies
    1. La projection de ce matin n'a pas inclus le sondage CROP. Il a été mis à jour cet après-midi.

      Delete
    2. Alors quel(s) sondage(s) comprenait la projection du 30 août au matin? La projection qui inclut le CROP présente le plus haut taux de votes pour QS ("In fourth is Québec Solidaire, at their highest point in the campaign with 7.8% support.").

      Comment expliquez-vous la baisse de Québec solidaire dans Laurier-Dorion avec cette hausse des intentions de vote au niveau national, dans le dernier CROP?

      Delete
    3. "Alors quel(s) sondage(s) comprenait la projection du 30 août au matin?"

      Forum.

      "Comment expliquez-vous la baisse de Québec solidaire dans Laurier-Dorion avec cette hausse des intentions de vote au niveau national, dans le dernier CROP?"

      La projection est régionale. Le résultat pour QS sur l'île de Montréal dans le sondage CROP était inférieur à celui (estimé) du Forum.

      Delete
  11. Est-ce que vous avez une version qui est facilement copiable (ctrl+c) pour mettre dans un tableur excel?

    ReplyDelete
  12. I noticed that the number of undecided in both the CROP poll and the SEGMA regional polls is very high: 19% of Quebec voters are undecided or won't say according to CROP. This is something worth noting as it makes this whole election a bit of a crap shoot.

    In Brome-Missisquoi (my riding) that undecided/won't-say number runs to 31% (higher than the support for any party in a tight 3-way race). Yet the difference between high and low of possible outcomes in your model projection for Brome-Missisquoi is 2-3%. I would think that the uncertainty here would be higher. What is the probability of votes for a given party falling within the high low range assigned to each riding?

    ReplyDelete

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.