Monday, April 1, 2013

Incentive for Horwath to defeat government drops

Over the long weekend, a new Forum poll was published by The Toronto Star, showing that the race in Ontario remains close. But the Progressive Conservatives and Liberals appear to be widening the gap between themselves and the New Democrats, decreasing the incentive Andrea Horwath might have had to defeat Kathleen Wynne's minority government.

Methodological note: the report from Forum Research is the usual, with plenty of cross-tabs but lacking unweighted sample data. Forum is not alone in this, as only Abacus Data and Ipsos-Reid routinely release unweighted data with their polls. Some other pollsters release the data on request, but in general disclosure is rather limited in Canada.

That makes it perhaps unfair to single out Forum, but they are one of the most active firms and IVR polling is, generally speaking, more susceptible to distorted sampling. We've seen this before - throughout the Quebec election campaign, telephone and online pollsters Léger Marketing and CROP had more stable results than Forum's IVR numbers, which varied more widely. That makes it all the more necessary to have this unweighted data, to help determine whether any fluctuations are real or not. But other firms need to up their disclosure as well. Even some of the most respected polling firms in Canada fail to release data as basic as regional sample sizes.
Forum was last in the field provincially between Feb. 26 and Mar. 1. Since then, there have been no statistically significant changes in support. The PCs picked up three points to reach 35%, the Liberals increased one to 33%, and the NDP was down three to 26%.

The trend is important, however, as the gap is widening between the two larger parties and the New Democrats. And the NDP has placed third in six of the seven polls that have been released since Wynne became Liberal leader.

There is a hidden advantage in these numbers for the Tories, as they were ahead by six points or more among Ontarians aged 45 and older. There is still a gender gap, however, as the PCs were first among men (by a wide margin) but third among women.

Regionally, the Progressive Conservatives led in eastern Ontario with 45% (+13), followed by the OLP at 27% and the NDP at 20% (-11). The Tories were also ahead in the 905 area code (38% to 32% for the OLP and 25% for the NDP) and in southwestern Ontario (38% to 30% for the NDP and 27% for the OLP).

The Liberals led in Toronto with 46% support, well ahead of the PCs at 25% and the NDP at 24%. The Liberals were also narrowly ahead in northern Ontario, with 33% to 30% apiece for the Tories and NDP (note that in Forum's poll, northern Ontario stretches all the way south to Lake Simcoe).
With these regional numbers, the Progressive Conservatives would win 47 seats to 40 for the Liberals and 20 for the New Democrats.

Forum's seat projection model gives the Liberals the plurality of seats (they lose three) while the NDP drops to only 14, despite being several points up on their 2011 election result. Regional distribution cannot be blamed, as the NDP improved upon their 2011 numbers in every region except Toronto, where the Liberals were also down and the PCs were up (but uselessly so). I don't understand how it is possible for them to have these sorts of numbers, and if the Star knows they should tell the rest of us. With these counter-intuitive results, they probably shouldn't publish them if they don't know.

The personal approval ratings for the three leaders look very good for Wynne and Horwath, but deceptively so. Wynne's approval rating was up six points to 40%, while her disapproval rating stood at 34%. Horwath's approval rating was steady at 44%, but her disapproval rating increased by five points to 30% - its highest level since June 2012. But compared to Tim Hudak's numbers, Wynne and Horwath are doing very well. Hudak's approval rating is only 27%, with a disapproval rating of 52%.

Why these numbers are deceptive, though, is due to the other questions Forum asked. When it comes to which leader shares respondents' personal values, Hudak and Horwath tied for 22% with Wynne at 21%. On who respects democracy, Horwath narrowly edged out Hudak with 22% to 20%, with Wynne at 18%. And on who is best able to handle the economy, Hudak was ahead of Wynne with 24% to 23%. Horwath was well behind at 15%.

This would seem to suggest that while Ontarians like Wynne and Horwath a lot more than Hudak, they don't consider them better leaders. That is a silver lining for Hudak, particularly his good showing on the economy.

But the PCs have made it clear they want to defeat the government, so Hudak might not be looking for a silver lining. The power to send Ontarians to the polls is in Horwath's hands, and with these numbers she has little incentive to do so.


  1. Sorry to nitpick, Éric, but unless I'm mistaken Forum's last poll had the PC's and Libs tied at 32%, meaning you might want to edit the following passage: "... the Liberals dropped one to 33%". Then again, I also noticed that in Forum's report for this new poll, the text states that the Liberals are at 32% whereas the data tables all show 33%. Confusing!


    1. Whoops, thanks. I had it as +1 in my notes, so just a mental lapse.

      Whenever there is discrepancy between sets of numbers, I always use what is in the tables.

  2. As I've said elsewhere ( ) I have a hard time with regional polling based on such small samples. The margin of error is about 6.5% in Forum's poll. Seeing ridings Sudbury stay Liberal, with no incumbent, just seem to go against the zeitgeist.

    1. The point is well-taken, but the regional polling nevertheless quite consistent. The MOE is important, but the most probable reality is closer to the poll's results than the edge of the MOE.

  3. There must be a bias against the NDP if these pollsters keep undervaluing the projected seats so much. Why not be transparent about how seats are allocated?

    1. That's a mighty big assumption - Forum has never demonstrated any sort of bias against the NDP. Watch what you say.

      I believe that Forum's projections are based literally upon the respondents they have in a riding. Obviously this is an issue given that the margin of error would be so large as to make them useless. This is what I believe they do.

      On the other hand, it is always the Liberals that end up far with more seats than they should have based on the numbers given, and never the PCs or NDP. That makes me think there may be something more to what they do, some sort of tweaking or weighting they give to maybe balance things out. This doesn't imply "bias," but that their methodology for these projections is different from what Eric, I, or others use. Like Eric said, its important that either the Star or Forum explain.

    2. I don't think Forum bases their projections on riding results, as I was told they use a "regional shift" model. Also, if they did use that method, they would have far more fluctuation in their results. Instead, they are rather consistently below what both you and I and LISPOP show for the NDP in Ontario.

    3. And, statistically speaking, their methodology could have a bias against the NDP, just an unintentional one.

    4. Hm, so much for that hypothesis! I'm very curious what kind of numbers they're using in a regional shift model that could produce numbers like this so steadily.

      And yes, as you said it could have an unintentional bias against the NDP - but at some point they need to look your site or LISPOPs and assume something must be up with what they're doing.

    5. I just find it disappointing that the Star reports them without question. The best that the NDP ever did in Forum's seat projection was 27 seats, and that was at 35% with a three-point lead over the PCs and an eight-point lead over the OLP. Still, the PCs and OLP won 40 seats apiece with those numbers.

  4. Which seat do the conservative hold in Atlantic Canada?

  5. The three parties in Ontario have been bouncing around within a relatively narrow range for the past couple of months, i don't think a single poll is going to make or break anyone's willingness to have an election. I think the ONDP wants more time to raise money and recruit candidates and concentrate on the upcoming byelections in Windsor and London...their willingness to bring down the government is not going to change based on whether this weeks poll has them at 26% or at 29% or whatever.


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.

Note: Only a member of this blog may post a comment.