Wednesday, August 22, 2012

Rogue poll or a turning of the tide for PLQ?

So you might have heard that a new poll came out today. A collective scoff emerged from Quebec when The National Post released the details of the newest Forum Research poll. The survey puts the Liberals ahead of the Parti Québécois by six points, after having the PLQ trail by four only a week ago. Is it a rogue poll? A 19 out of 20 result? Just a badly done bit of public opinion polling? Or the first sign that things are turning around for Jean Charest? We'll find out soon enough.

I wrote about the poll for The Globe and Mail this morning, so I invite you to read the article for more detail on how it stands up to other surveys in the province.

The projection has swung hard because of this poll, with the Liberals up 4.3 points to 34.2% and the Parti Québécois down 3.5 points to 31.5%. The CAQ, with 23%, is down 1.6 points. Québec Solidare picked up 1.8 points to hit 7.5%.

With the Liberals now in front in the vote projection, they also move ahead in the seat projection. They picked up 20 seats to surge to 58, five short of a majority, while the PQ dropped 16 seats to 49. The CAQ dropped five seats to 16 and Québec Solidaire increased their share from one to two.

It goes without saying that the poll has injected a great deal of uncertainty into the projection, which now gives the Liberals and the PQ the possibility of forming a majority government or winning as few as 35 seats. In other words, because of this poll the projection model has no idea what would happen if an election took place today. This was the same sort of thing that occurred when Forum released their eleventh hour poll on the eve of the Alberta election, suggesting that a swing from Wildrose to the Progressive Conservatives was taking place. That a swing was underway turned out to be right, but it was the kind of swing that people were expecting to happen in Alberta but never seemed to (at least in the last week of polls). This swing from the PQ to the Liberals is completely unexpected.

For that reason, I don't want to get too much into the details of the new projection. If this is a rogue poll, it will swing back hard when Léger Marketing and CROP release their new numbers. If it isn't, and the two Quebec-based firms show the same Liberal surge in support, then we will know something is really going on. We should probably give Forum the benefit of the doubt for now and consider them innocent until proven guilty. Few believed the first poll that showed the New Democrats surging in Quebec in the 2011 federal election, after all.

After watching the three debates that have taken place so far, it does not seem unlikely that Jean Charest has indeed stopped the bleeding. I'm not convinced, however, that he did enough to cause such a huge shift in voting intentions. The final debate between Pauline Marois and François Legault, which takes place tonight, will be very important in deciding how the anti-Charest francophone vote will go. Marois has the advantage (even in Forum's poll), so the onus is on the CAQ leader to perform.
Forum was last in the field on August 13, and since then the Liberals increased their support by four points to lead with 35%. The PQ dropped six points to 29%. The CAQ was down one point to 24%.

Likely due to Françoise David's good performance in the debate, QS increased their support by three points. At 9%, the party is at their high watermark of the campaign so far. But Forum also added Option Nationale to their polling. At 1% they are not a big factor, but combined with the three points that QS gained we start to see where the PQ lost their six points.

But the Parti Québécois does still have an ace up their sleeve, as Forum found that 82% of their voters are very or somewhat enthusiastic about voting for the party. By contrast, enthusiasm sits at 72% among Liberals and 70% among CAQ voters. If that's a proxy for turnout, that transforms a six-point Liberal lead into a neck-and-neck race with the PQ.

On the other hand, 48% of PQ supporters felt that Françoise David won the Sunday debate, compared to 23% who said Marois won it. CAQ voters were more split between David and Legault, while PLQ voters felt Charest won. This indicates a less than strong attachment to Marois from PQ supporters.

Riding polls

Another riding poll for Nicolet-Bécancour has emerged. The riding is looking like one of the most interesting in Quebec, as a poll by Cara-Telecom for Le Courrier Sud and CKBN Radio sees a four-way race. With only 223 surveyed the margin of error is quite high (+/- 6.6%), but the poll gives Jean-Martin Aussant 29.7% support to 26.2% for the CAQ's Donald Martel. The Liberals are not far behind at 24.1% while the PQ brings up the rear at 20%.

The results are not very different to Cible Recherche's poll taken over some of the same days, which also put the race as one between Aussant and Martel. The only difference is in Liberal and PQ support, but the margin of error takes care of that. Nicolet-Bécancour will certainly be a riding to watch on Sept. 4, and if the result of the election is a minority government that one seat could be extremely important.


  1. How close do you have the QS in Laurier and in Sainte-Marie?

    1. Riding level projections can be found here:

  2. So one poll and you change everything? Back in 2011, you used to average polls from like 10 months ago, meaning that you never captured the NDP wave, but now you only include the very last (and weird) poll? lol

    1. I've done six provincial elections since the last federal election using the current model, and it has performed well.

      As I've explained since the federal election, I made changes to the date weighting so as to be able to capture those changing trends. The federal projection would have been much better had it used the current weighting system.

      And no, I don't only include the last poll. The polls included in the projection are at the bottom of this page. Forum's poll being the newest makes it the heaviest in the projection.

  3. Do you think any of this is a backlash over the whole "no non-Christian religious symbols allowed" nonsense?

    I haven't been following La Presse as much as I normally do and I realize the English stuff I read has a big selection bias, but Marois seems to be putting even the "money and the ethnic vote" nonsense to shame.

  4. This comment has been removed by the author.

  5. I love the work you do and would like to thank you for sharing it publicly. I am wondering if you have ever considered tweaking your model for polls taken in Québec (either provincially or federally) to capture Quebecois' behaviour of voting for non-traditional parties as a way to punish or get a change (example, ADQ opposition and NDP wave) by giving more weight to recent trends than you would for polls taken in other jurisdictions.

  6. I have noticed that the Forum polls have had the most extreme results in the election.

    Of the highest and lowest results for the four largest parties, six of the eight values are from Forum when only 3-4 should be.
    Highest PQ - 39% Forum August 1st
    Lowest PQ - 29% Forum August 20th
    Highest Liberal - 38% Forum August 1st
    CAQ lowest - 14% Forum August 1st
    QS highest - 9% Forum August 20th
    QS lowest - 4% Forum August 1st

    Also Forum seems to be trending high for the Liberals and PQ when compared to CROP and Leger

    Combining the PQ and Liberal results is interesting because of the order of the pollsters:
    77% - Forum Aug 1st
    66% - Forum Aug 13th
    66% - Forum Aug 7th
    64% - Forum Aug 20th
    64% - Leger July 31st
    63% - Leger Aug 13th
    61% - Leger Aug 16th
    61% - CROP Aug 14th
    61% - CROP Aug 8th

  7. A shocking number in this Forum poll is 40% for the Liberals in the Quebec City RMR.

    I was under the impression that Québec City was a lock for the CAQ. If this poll isn't a "rogue", these additional seats for the Liberals might be enough to give them a minority government.

    Here's hoping that CROP and Leger will publish a new poll soon.

    Good work, Mr Grenier.

  8. Good day Eric

    Just suggesting that maybe you can make a faq section, as every so often there are people asking questions that you basically answered many thousands times already(such as the one from Kevin above). Such happening can not only be potentially frustrating for you, but tiring for viewers looking for more meaningful comments. Making a faq page might decrease the chance of such questions, & directing everyone who do asked to the faq can also not only save you time but save space in the comment sections.

    Once again, appreciation for you work.

  9. Guys dont panic. Several sources are saying this forum poll was flawed because almost everyone manning the phone spoke no French, and almost all contacts were initially made in English, ensuring that the sample is flawed because unilingual Francophones are still a big minority in Quebec. And many bilinguals still wont take English calls at home.

    1. Forum uses an IVR system, not live-callers, so there were no non-French speakers manning the phones (robots manned the phones). I was also told by Forum that the rumours that the calls started off in English are false.

    2. Anonymous,

      First, could you please point us towards some of these sources?

      Second, I really doubt Forum would be so foolish as to jeopardize its credibility and reputation by botching the quality and consistency of its methodology in such a way. Pollster conspiracy theories generally don't make sense.

      I'm as eager as everyone to see if the other pollsters will confirm or contradict this trend, but generally speaking it is not in the interest of the credibility of a pollster to appear totally out of whack with what the others are finding or make a prediction that turns out totally wrong come election day. Therefore, any attempt to "manipulate" or any lack or rigor represents a considerable risk. I'm pretty sure Éric would agree.


    3. Why would we "panic" ?

    4. Eric and Dom, I'm almost certain I've been call by Forum. It was an automatic message, in english, starting by "This is a message from Forum research. You have been carefully selected...". Hard to say the rest of the message, as I was calling name to the bot (in french).

  10. If the shift observed by Forum is reflected in other polls, I would hazard a guess that it could represent scepticism on the part of Quebecers, concerning the PQ's "Secular Charter", which was receiving significant media attention around the time that Forum was in the field.

  11. It's funny in the US elections last for two years and in Canada they last for 2 weeks. People are just starting to pay attention now.

  12. What would the numbers need to most likely be for QS to gain a third seat (or more)? Where would those most likely be?

  13. This used to happen in the 1990s: Leger and CROP would show margins of 5-10% for the Bloc or PQ, then one poll would show a Liberal lead which would get dismissed by everyone, and then on election day the Liberals would tie or win the popular vote (1994 Provincial, 1997 federal, 1998 provincial, 2000 federal).

    Federalists tend to be older and wealthier and therefore have higher voter turnout rates. In low turnout elections, this can send the polls askew. Where 50-60% vote, turnout is a big factor. The main outlier in the 1990s was the 1995 referendum, which the polls got exactly right because 95% of the population voted.

    The Liberal bounce has been less evident federally since the 2000s because of the revival of the Conservatives and NDP (in the 90s, Reform and NDP got 2% or less of the vote in Quebec).

    The provincial polls got more accurate once CROP and Leger started adjusting their "repartage des indecis" to automatically add 2-3% for the Liberals to every poll.

    I don't know if that's what is going on here, but it's possible.

    1. Except that since 2004, the Liberal support has been polled accurately by Léger (and to a lesser extent by CROP) for all elections in Québec. If anything, they have been overestimated. In 2007, Léger gave the Liberals 35%, Crop, 34%. They ended up with 32%. In 2008, Léger and CROP gave them 45%, they got 42%.

      In federal elections in Québec, it's the same. In 2006, Léger gave them 33%, they ended up with 33,9%. In 2006, Léger gave them 20%, they ended up with 20,7%. In 2008, Léger gave them 24%, they ended up with 23,7%. In 2011, Léger gave them 15%, they got 14,2%.

      The last time the Liberals outperformed the polls was 2003, since then they have underperformed them slightly. And Léger Marketing has been spot on in federal elections (more or less 1 percent) and has slightly overestimated them in provincial elections (2 or 3 points more).

      This time, Liberal voters are very worn out. The Liberals lost Argenteuil and there is no real enthusiasm for the party on the ground. Even anglos have started talking of defecting to the CAQ. I highly doubt that they will outperform the polls significantly.

  14. Thank you for the constant polling update with related projections, it is a joy to watch the election unfold one poll at a time. May I suggest complete riding projections as were done for the last federal elections, it was interesting to see the swings and projections on a riding per riding chart

    1. I will get this one for you Eric.

      They exist, just click on the article itself.

  15. It would be nice to see a full projection of riding per riding polling as was done during the last federal elections. I am especially curious with inroads made by QS in PQ strongholds in montreal, as well as the Nicolet riding where Aussant is showing surprising strength. Good work, keep it coming

    1. Riding-level projections are being maintained on the site. Click on the chart at the top of the page (as indicated), or go here:

  16. Bonjour,

    May seem odd and not really a question, but I'm intrigued to know, are our cellular phone included in the random phone number draw ?

    A whole lot of people asked for it in my entourage.


    1. Hi

      I only have a cellphone and I get calls from pollsters asking me who I will vote for.

  17. Bonjour Éric,

    Thanks again for this interesting post. I am wondering, in case of a CAQ surge in the polls, what would be the first riding to fall for the CAQ?

    Bourassa-Sauvé? Anjou? Marquette? Saint-Laurent? Verdun? Pointe-aux-Trembles?

    I noticed that in your projections, the CAQ has a maximum of one gain in the island of Montreal, what would it be?



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.