Wednesday, May 21, 2014

Abacus shows tie in Ontario, PC edge among likely voters

After the back-and-forth see-sawing of the polls since the start of the provincial campaign in Ontario, it is comforting to see a poll with results that more comfortably fit into what the other pollsters have been saying. Whereas some polls see a PC lead and others see a Liberal lead, the new poll by Abacus Data for the Sun News Network splits the difference, putting the Liberals and PCs in a tie. But then the poll also agrees with what Ipsos Reid has been showing as well - that the Tories have an advantage among likely voters.

The results from Abacus differed only a little from the previous projection, so this is the first update in some time without a dramatic swing in one direction or the other. The PCs retain their lead at 35.8% (or between 34% and 39%), while the Liberals dropped slightly to 34.3% (or between 33% and 38%). The New Democrats were up to 23.4% (or between 22% and 25%), while the Greens held steady with 5.2% (or between 4% and 7%).

The Liberals also remain ahead in the seat count, with 44 to 41 for the PCs. But the NDP has rebounded, back to 22 seats.

While the precise seat estimate favours the Liberals, the ranges favour the Tories. They are projected to take between 38 and 54 seats, just grazing the minimum needed for a majority. The Liberals sit at between 32 and 52 seats, while the NDP is at 16 to 23 seats.

Recall that these are just the most likely outcomes. There is a 9% chance that the Liberal total will fall between the High to Maximum level (52 to 64 seats). That increases to 14% for the PCs (54 to 67 seats).

Regionally, the PC numbers have fallen back down to earth in Toronto, primarily to the benefit of the NDP. The Liberals have been steadily dropping in the 905 area code, while they have spiked in eastern Ontario. The NDP have been making gains in southwestern Ontario, and northern and central Ontario remains a three-way mess.

We have not heard from Abacus at the provincial level in Ontario since August 2013, so there are no trends to look at.

The PCs and Liberals were tied in the poll among all eligible voters with 33%, while the NDP trailed not too far behind at 26%. The Greens were at 6% and 2% of Ontarians said they would vote for another party. Of the total sample, 14% were undecided.

Among likely voters (determined by Abacus according to how respondents answer a half-dozen questions related to likelihood of voting and attention paid to the campaign), the PCs moved ahead with 36%, while the Liberals were unchanged at 33%. The NDP and Greens each dropped a point to 25% and 5%, respectively. Note that the projection model uses the likely voter numbers.

Among eligible voters, the Liberals led in Toronto with 40% to 27% for the NDP and 26% for the PCs, while they were also ahead in eastern Ontario with 42% to 31% for the PCs and 20% for the NDP.

The Tories were in front in southwestern Ontario with 38% to 30% for the NDP and 23% for the Liberals. The PCs also edged out the Liberals in the GTA/Hamilton-Niagara region with 37% to 33%, with the NDP at 23%.

The New Democrats led in northern Ontario with 37%, well ahead of the Liberals at 28% and the PCs at 27%.

Abacus defines its regions by postal code, so M is Toronto, L is the GTA/Hamilton-Niagara region, K is the east, R is the southwest, and P is the north.

The poll had oodles of information, and I invite you to check it out. What I found most interesting is how Abacus sliced up the electorate. It found 16% of Ontarians to be 'core PCs', with 10% being 'core Liberals' and another 10% being 'core NDP'. That's the base.

Swing voters represent the rest. The largest group are Liberal-NDP swing voters, representing 21% of the electorate. This partly explains the Liberals move to the left. Another 12% are PC-NDP swing voters, while just 10% are Liberal-PC swing voters. Another 13% are complete swing voters, willing to vote for all three parties. Things are not always as simple as the left-right spectrum. This is something that Abacus will be tracking, and it will be interesting to see how these groups move over the next few weeks.

20 comments:

  1. "The results from Abacus differed only a little from the previous projection, "

    In other words the "jitter" I mentioned before. Which I think will be a feature of the scene until we have the actual election.

    ReplyDelete
    Replies
    1. @ Peter

      There is no jitter as you say. There is simply different polling companies with different methodologies, results and possibly agendas.

      The polling firms have been extremely consistent (close to the point of saying the voters are locked in) since the start of the 2014.

      to illustrate this I have taken the CPC support as reported by the individual polling firms since the start of the year:

      EKOS 3 polls CPC support 27.4, 30.3, 31.6

      IPSOS Ried 4 polls CPC support 34, 37, 39, 37

      Forum 5 polls CPC support: 38, 32, 35, 36, 35, 38

      Innovative research 4 polls CPC support : 30, 33, 32.8, 32.8

      The polling firms are basically saying that CPC support has not changed since the start of the year. They are basically saying that the other polling firms are doing something consistently wrong to show different results.





      Delete
    2. BC That is precisely "jitter"

      In other words each pollster is consistent so the results depend on the pollster not the public. You can say "No Change" and be right on with that pollsters previous poll.

      Delete
  2. Using the reported numbers, I get:

    43 PC
    42 OLP
    22 NDP

    With the likely-voters, I get:

    45 PC
    42 OLP
    20 NDP

    With the average, I get:

    45 PC
    44 OLP
    18 NDP

    This is becoming a very tight race. The seat count changes with every slight percentage point change and turnout could "make or break" a party on election night.

    ReplyDelete
  3. This is the first poll I actually find plausible.

    ReplyDelete
    Replies
    1. People knock Abacus a lot just because they poll for Sun News, but frankly I find them to be exemplary among Canadian pollsters for their transparency and professionalism. And now with verteran pollster Bruce Anderson on board as chairman, I'd say their competence and integrity are definitely beyond reproach.

      Dom

      Delete
    2. I don't want to knock Bruce Anderson who is a very fine man but, he does have family connections to the federal Liberal party. I don't believe those ties have influenced his professionalism.

      Delete
  4. Looking back at 2011 - Abacus, FORUM and Ekos had the closest numbers to the actual results, with Ipsos - while blasting the others - having the worst. ( Abacus had the final 3 point Liberal victory spot on, while Ipsos had the Liberals winning by 10).

    Point being, different firms, different methods. But what is becoming very apparent - some methods are much more reliable/accurate than others. Perhaps, it's time Eric takes THAT into consideration.

    ReplyDelete
    Replies
    1. Track record is taken into account.

      Delete
  5. Great stuff Eric! Glad to see one poll showing a dead heat and at least partially validating the numbers being put up here. Will you be adding the riding polls done by Forum in Trinity-Spadina and Scarborough to the byelection barometer?

    ReplyDelete
    Replies
    1. Yes, I update the barometer on Fridays.

      Delete
  6. the ranges in the projections seem strange. the low is not as low as the high end. take the liberal 34.3-32.9=1.4 while 37.7-34.4=3.3. So, like, why is the "minus" smaller than the "plus"? is it an artifact, or something real?

    ReplyDelete
    Replies
    1. The ranges aren't symmetrical, they are based on how the polls have under- and over-estimated support of a party depending on its role in the out-going legislature.

      So, this means that the model assumes that the polls will under-estimate the support of the governing party more than they will over-estimate it.

      Delete
    2. Thanks that helps, so basically for the LIB & PC the "+" is bigger than the "-" (for % support), while for the NDP & Green it is closer to even. Now that I think about it a bit more I see how it could makes sense.
      It's funny how this "room to grow/shrink" in the predicted support numbers translates non-linearly into the seats projection. The PCs seem to have a lot more room to grow in the seat projection +26 vs -14 and the Liberals have more room to shrink +24 vs - 20, same with the more room to shrink NDP +5 vs - 11. This translation from support to seats range (especially the NDP) does not seem intuitive to me, given the way popular support typically translates to seats in a FPTP system.

      Delete
  7. The best option for the Federal Tories is a continuation of the Liberal minority government, going into 2015.

    ReplyDelete
    Replies
    1. It's the best option, but not by a long shot. The federal Liberals can still form government in 2015 despite who is in power in Queen's Park.

      Federal and provincal PC governments have co-existed in the past. Martin and McGuinty were in power during the same time. Even in 2006, when Harper became PM, the federal Liberals still won the plurality of seats despite the so called Liberal baggage at Queen's Park.

      Delete
  8. Todays Toronto Star

    "In the latest Forum Research survey, Kathleen Wynne’s Liberals lead with 41 per cent to 34 per cent for Tim Hudak’s Conservatives, and 20 per cent for Andrea Horwath’s New Democrats. The Greens, led by Mike Schreiner, have 4 per cent."

    ReplyDelete
  9. New Forum poll:

    http://www.thestar.com/news/queenspark/2014/05/22/poll_tory_cuts_and_lowkey_ndp_campaign_helping_liberals.html

    ReplyDelete
    Replies
    1. That gives me:

      60 OLP
      33 PC
      14 NDP

      That is not all that far from Forum's own numbers. I can't say I trust Forum's simulator as much as Eric's, but at least, it somewhat proves that my own methodology isn't all that crazy since a company out there seems to use something somewhat similar.

      Delete
  10. where can I see the poll results from 308 ... on a weekly basic since the Ontario elections where called? please help... I want the above chart ... however I forgot to save the progression of the stats ... I need this for my class... thank you!

    Paul ;-)
    please emailme @ guardpol@hotmail.com

    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.