Friday, March 29, 2013

Forum Ontario Poll: 35% PC, 33% OLP, 26% NDP

What a beautiful day! And here I am, stuck inside reading a new poll for Ontario's provincial scene, out by everyone's favourite pollster. The things I do for you people.

Forum Research (Ont. Provincial - March 26-27, 2013 - +/-3%)
Prog. Conservative: 35% (+3%) - 48 seats (+10 seats)
Ontario Liberal Party: 33% (+1%) - 38 seats (-3 seats)
New Democratic: 26% (-3%) - 21 seats (-7 seats)
Green Party: 5% (=)

Compared to the last Forum poll out in February, both the PCs and the Liberals have gained while the NDP fell back to their lowest levels in a Forum poll since March 2012. The accompanying Toronto Star article for this poll had a great opening paragraph that may explain why: "The longer Liberal Premier Kathleen Wynne is in office the more Ontarians seem to like her, a Forum Research poll suggests."

However, not everything is sunshine and lollipops. While Forum's own seat projections give the Liberals a minority government, I wouldn't put too much stock into them - as always, Forum's seat numbers are seemingly based on what riding the respondents are in, and with an overall sample size of around 1,000, there's no way you can get a representative sample within all 107 ridings.

My own seat projection hands the PCs a minority government, with the Liberals not too far behind and the NDP sitting with a good amount as well. This seems like a result much more in line with what the poll is showing in its topline and its regions, 33% simply isn't enough for the Liberals to knock the PCs out when they're sitting above them.

Regionally, the Liberals lead in the GTA, 37% to 33% for the PCs. That lead is mostly built in Toronto, where the Liberals lead with 46% to the PC's 25% and the NDP's 24%; in the 905 Belt, the PCs lead 38% to 32% for the Liberals. At 25%, the NDP are a far cry from the 30%+ they got before in the 905, when the Liberals were collapsing across the province.

The PCs lead in Eastern Ontario, 45% to 27% Liberal, while the Liberals come out first in a three-way race in the North with 33% to 30% for both the PCs and NDP. Its a believable result, and this is where the majority of the NDP loss came from compared to February - that being said, Forum has had an odd history with its Northern Ontario numbers.

In Southwestern Ontario, the PCs lead 38% to the NDP's 30% and the Liberal's 27%. This is where our two expected by-elections will occur in London West and Windsor-Tecumseh - based on those numbers, the Liberals would win London West (39% to 30% NDP), but lose Windsor-Tecumseh (41% NDP to 33% Liberal).

Forum also did some approval/disapproval stuff, with good news for Kathleen Wynne, for whom 40% approve to 34% disapprove. Most of this comes from the Liberals (72% approve), but NDP respondents also gave Wynne a positive rating (41% approval to 30% disapproval).

Tim Hudak is still in trouble, however, with only 27% approval to 52% disapproval!! That's a huge number, McGuinty-esque really. Even a significant portion of PC supporters in this poll disapprove of Hudak (23% to 56% approval). He's absolutely hated among the other party supporters, except apparently among the Greens, of which 32% approve to 53% disapproval - but its a small sample.

For Andrea Horwath, 44% approve to 30% disapprove of her job as third party leader. She earns great scores among the NDP, but also among the Liberals (45% approval) and even the PCs (26% approval). Its a wonder how Horwath, who is immensely popular, can't bring her party up with her to beat the PCs, who have an immensely unpopular leader (but continue to lead anyways).

Forum offers a possible explanation, with further polling on questions such as "who shares your values the most," or "who is best to handle the economy." That latter one is important; 24% say Hudak is best to handle the economy, compared to 23% for Wynne. Horwath, despite her personal popularity, manages only 15%. Even when you take away don't knows or none, Horwath sits at only 22%.

Comparatively, 36% think Horwath's the one that cares the most about the most vulnerable among us - yet we all know which issue is the more pressing one in the minds of Ontarians, don't we.

Anyways, here's a map:


  1. This poll is somewhat disappointing. It seems odd that with improvement of Wynne's personal numbers and a drop in NDP support, Liberal support stayed flat and PC support actually went up. Why do you think this happened?

    Also, how did the riding of Brant manage to go NDP in your projection when the NDP only are projected to win 21 seats, three more than they now have, while Sudbury manages to stay Liberal even though the NDP came within two points of winning it last time?

    Finally, do you think that if these seat numbers actually happend that Wynne would hand over power to a PC minority government that would not have the votes to make the sweeping cuts and sweeping changes they are proposing or would Wynne try to hang on and govern as a minority government despite having fewer seats?

    1. As to your first point, there is still a lot of anti-Liberal sentiment out there. While I don't expect to see NDP supporters moving en masse to the PCs, there's nothing out there saying that former Liberals who have left for the PCs will come back. Its clear to me that most of the current Liberal rise is due to soft NDP support coming back to the Liberals. Only time will tell whether or not Wynne can win over the more centre-right people that have felt disenfranchised with the Liberals and move to the PCs, thats including those that left in the 2011 election as well as post-2011.

      To your second point, this poll is based on regional numbers. In the North, for instance, the NDP earned just shy of 38%; in this poll, they've dropped down to 30% across the region. Thus their numbers in Sudbury will go down, as the Liberal numbers go up (given that 33% is 3% more than they earned in 2011).

      In Brant's case, the NDP in this poll are sitting at 30% in Southwestern Ontario - a full 8.5% higher than they earned in 2011 in the region. The Liberals are down about four points to 27% in this poll. Accordingly, the NDP numbers climb up in Brant, while the Liberal numbers go down. That's why they can eek out a win.

      On your final point, it would really depend. The ideal analogue is the situation we have in Quebec, where the PQ won a minority government but clearly didn't earn much of a mandate. Even so, the Liberals allowed them to government and sit in opposition alongside the CAQ. There is really nothing saying the Liberals and CAQ can't form a government, as they're broadly similar parties with similar interests. Yet they don't, because the Liberals need to lick their wounds and its easier to simply be a thorn in the PQ's side, while simultaneously allowing them to accumulate a bad reputation while in government.

      I can see a similar situation in Ontario. In a minority situation, the Hudak PCs can't necessarily do too much, and what they will be allowed to get away with would probably only provide fodder for the Liberals and NDP. Much would also depend upon the leadership situation after an election loss. I'm not saying Wynne would be tossed out, or Horwath for that matter - but there's always a possibility.

  2. I continue to have a hard time giving too much credibility to the Forum polls as the make up of those surveyed fluctuates wildly. Last poll had a larger number of previous NDP supporter therefore a higher pro-NDP vote. This poll has a large number of CON voters and a much smaller number of NDP voters based on past vote history therefore CONs up and NDP down. I fail to see how Forum seeks to find a representative past voting base to determine changes since that election. Perhaps you can explain it for those who visit the site.

    1. I can't exactly speak to Forum's methodology or what they do with their samples, as I'm part of their organization. What I can tell you however is that almost every other pollster out there, whether Angus Reid or Leger, will have those kinds of weird and questionable sample size numbers - the most recent Leger Quebec poll had the PLQ sample size larger than the pequiste sample size, when obviously we know who won in 2012. So Forum isn't exactly an exception.

      What I tell you, or at least what I know, is that it is extremely hard for a polling company to find an exact sample in this day and age. You need to remember that probably less then 5% of people will ever receive and answer a call from a polling company, in fact some postulate its less than 1%. This isn't an exact science, they work with what they can get. Its very possible that this is all they could manage. If so, they'll then rely on the weighting and the margin of error to correct any imbalances. So long as they don't have egregious samples or weighting, they'll usually come out with sound results. This in no way constitutes as egregious.

    2. As I'm *not* part of their organization. If only I was, I do need a new job.

  3. Shouldn't the regional sub-samples' MOE be rather high? That's why I tend not to use them. If they were as rigidly-defined and well-polled as US States, I might put more stock in the numbers I see around the web for these "Sub-provinces".

    1. As I suspected, if you put the sample size and observed proportion of the NDP's northern vote into this form (213, 0.3): ( ) you get an MOE of +/- 6.3%.

      The Libs could be at 26.7%; the NDP could be at 36.3% in the North! ;0

      Your model needs a way to predict a generic Conservative vote (and the concomitant swing) for André Arthur's old seat. I haven't heard that he's running again. And the Conservatives (and their tag-alongs) have really irritated middle QC since 2011, so he might not even have the same allure as in 2006.

    2. It depends on the region and the poll, but yes their MOEs are fairly high, in the range of 5% depending upon the region. However, I trust that an average of these regional numbers will give you a fairly steady and accurate picture of the poll's trends over time. Luckily we have a lot of data, at least monthly, to build that trend on. Its when you get to situations like, say, Nova Scotia, where polling is done every quarter - much less data to work with, and more likely that the MOE's quirks will come into play.

    3. Ah, yes, Arthur's seat. You're probably right that he won't be running again - from what I hear he was a crap MP anyways. I'll probably do something with Portneuf-JC when the final redistribution report is out and we have the official transposition. I'm not sure what a generic Conservative would look like in the riding; they had a candidate in 2004 and 2006 that won between 21-23%, so its a relatively strong riding for them, Arthur or no Arthur.