Monday, March 19, 2012

Tight federal race in British Columbia

Two new polls show that the race in British Columbia is a close one between the federal Conservatives and the New Democrats, while Stephen Harper's party has gained nationwide since late February.
EKOS Research was last in the field for iPolitics.ca between February 21-28, and since then the Conservatives have picked up 3.9 points to lead with 35.1% nationally. The New Democrats are up 0.5 points to 29.7%, while the Liberals are down 2.1 points to only 19.6% support.

The Greens are down 0.2 points to 8.1%, while the Bloc Québécois is also down 0.2 points to 5.8%.

EKOS thus joins some of the other recent polls that have put Liberal support back around 20% with the Conservatives in the mid-30s. I suspect the NDP leadership race, which comes to a close this weekend, might change things again.

In British Columbia, the Conservatives are up two points to 35.3%, narrowly edging out the New Democrats. They trail with 33.2%, down 3.3 points. The Liberals are well behind with 16.3% support, down 1.1 points since the end of February. The Greens are doing well here with 14.3%, up 3.3 points.

Justason Market Intelligence also released their latest numbers on the federal scene in the province. Their poll, conducted from February 24 to March 7, so only overlapping with EKOS's survey on March 6 and 7, found the New Democrats leading with 40% support. The Conservatives placed second with 30% while the Liberals were in third with 20%. The Greens brought up the rear with 8%.

Considering the margin of error, these two polls show that the race is very tight in British Columbia. It is clearly a contest between the New Democrats and the Conservatives, who seem to trade the lead with every poll: the NDP has led in seven of the last 15 polls stretching back to December, with the Conservatives leading in the other eight.

British Columbia will be an important battleground in 2015, as its seat allotment increases to 42. If Justason's numbers were repeated on election day, the New Democrats would take 18 of the province's current 36 seats, with 11 going to the Conservatives, six to the Liberals, and one to the Greens. With 42 seats, the NDP could win 21 to the Tories' 13 and the Liberals seven.

If EKOS's numbers were the election's results, the Conservatives would win 18 seats to 13 for the NDP, four for the Liberals, and one for the Greens on the current 36-seat map. With 42 seats, the Conservatives could win 21 to 15 for the NDP and five for the Liberals.

For the New Democrats, that means an extra three to nine seats. For the Conservatives, that means maintaining their current number or losing eight B.C. MPs. It might not be earth-shattering, but if it is a close election it could be the difference between a majority and a minority.

Elsewhere in EKOS's polling, the Conservatives lead with 34.6% in Ontario (+1.4), 61.2% in Alberta (+7.6), 31.9% in Atlantic Canada (+4.5), and 45.9% in the Prairies (+7.6).

The New Democrats lead in Quebec with 30.6%, up 2.3 points since the end of February. The party is also running second in Ontario (31.0%, +3.0), Alberta (18.6%, -1.8), and the Prairies (39.2%, +4.4). The Liberals are running second with 26.1% in Atlantic Canada (-2.6) while the Bloc Québécois is second in Quebec with 24.5% (-0.5).

With EKOS's results, the Conservatives win 147 seats on the current 308-seat map, with 96 seats going to the New Democrats, 52 to the Liberals, 12 to the Bloc Québécois, and one to the Greens. The NDP and Liberals could combine for a total of 148 seats, but that is seven short of a majority.

The Conservatives win 18 seats in British Columbia, 27 in Alberta, 16 in the Prairies, 54 in Ontario, 17 in Quebec, 14 in Atlantic Canada, and one in the north. They could win 165 seats on the 338-seat map, increasing their share from 47.7% to 48.8% of all seats.

The New Democrats win 13 seats in British Columbia, one in Alberta, 10 in the Prairies, 27 in Ontario, 39 in Quebec, five in Atlantic Canada, and one in the north. They'd likely win 104 seats in the expanded House, their share decreasing from 31.2% to 30.8%.

The Liberals win four seats in British Columbia, two in the Prairies, 25 in Ontario, seven in Quebec, 13 in Atlantic Canada, and one in the north. With a 338-seat House of Commons, the Liberals might win 56 seats, their share decreasing from 16.9% to 16.6%.

The Bloc Québécois wins its 12 seats in Quebec, of course, while the Greens win their one seat in British Columbia.

Using Justason's B.C. results, the national seat total would be 140 Conservatives, 101 NDP, 54 Liberals, 12 BQ, and one Green. This is a significant difference, since the NDP and Liberals could now combine for 155 seats, the bare minimum for a majority. On the 338-map, however, the result would be 157 Conservatives to 110 NDP and 58 Liberals. The combination of NDP and Liberal seats would be short of a majority by one, requiring the Greens to come on board.

This is a nice demonstration of the potential importance of British Columbia in an election that resembles EKOS's forecast. But other things would be at play: those 12 Bloc seats cause more trouble to the NDP than British Columbia's close contest, while maintaining their high support in Ontario and the Prairies would be absolutely necessary. And, in this case, the Conservative result of 22.9% in Quebec puts them in the running to hold on to power, particularly if the party does not do well in a province like British Columbia. Every seat might count.

28 comments:

  1. Eric, don't you think that if this Ekos poll actually was replicate din an election and the NDP was within 45 of the Tories with 31% compared to the Tories 35% - the seat gap between the Tories and the NDP would be a lot narrower than 54-27?

    ReplyDelete
    Replies
    1. You're talking about Ontario, I assume? I don't think the NDP vote is efficient enough.

      Delete
    2. The NDP vote in Ontario might seem "inefficient" at the 26% level and distribution we saw in the '11 election. But who knows where they go if they get into the low 30s. Who would have predicted the NDP surges in Scarborough-Rouge River and Bramalea-Gore-Malton last time.

      Delete
    3. True, but it tends to even out as they would under-surge in other regions.

      With the real Ontario-wide vote result, my model would not have had the NDP doing well in those two ridings, but they still would have been forecast to win 22 seats in the province.

      Delete
  2. BC seems to be the only place EKOS agrees with the other pollsters...

    ReplyDelete
    Replies
    1. That's appropriate. BC's trends never match the rest of the country.

      Delete
  3. Only 7 seats for the Liberals in Quebec? Seems off. Although, as I've mentioned before, EKOS in particular has usually given lower numbers for the Liberals than it would appear, so perhaps, given the poll, shouldn't be surprising.

    Also interesting that the Bloc would only win 12 seats, considering that they've been pegged to win between 20 or 40 seats in recent weeks, so that also seems a bit odd. It could be that Quebec is now basically assuming that Thomas Mulcair will become NDP leader, but it still seems off.

    Also, what do you expect the seat ranges would be? Is their a chance that the NDP could be reduced to third place or that they could be just seats apart from the Conservatives?

    ReplyDelete
    Replies
    1. Well, if the Liberals are at 15.1% in Quebec I don't see how they could win much more than seven seats.

      Delete
    2. Well yeah I understand that, but what I find odd is that EKOS has them at that level of support and 4th in Quebec when they've been polling at least 10% higher in the last few weeks. Their numbers for Quebec definitely seem off.

      Delete
    3. Off? How do we even know? The whole point of polls is a guess as to what the "on" number really is. No pollster ever gets it 100% "correct".

      Delete
    4. What I mean is it's hard to see the Liberals dropping 10% in Quebec within a week.

      Delete
  4. Eric,

    Do you know whether Ekos applied their "likely voter" screen to these results? Looking at the release, I can't tell.

    ReplyDelete
  5. How is the headline for this poll not "Ekos has Tories polling higher than the were going into the 2011 election".


    According to Ekos the CPC have 35.1% of the popular vote. This is an increase from the 33.9 and 34.6 that EKOS had the CPC at May 1 and April 30th, 2011 in the last poll they published prior to the vote where the CPC actually got 39.6.


    This EKOS poll actually shows the CPC support growing by 1.2%. In the real world this pushes them to 40.8% and an even larger majority.... not even taking into account the increase efficency of the CPC vote due to the extra seats.

    The 20% for the Liberals shows them losing 1% from what EKOS hasd them going into the last election, EKOS show the NDP falling off to 29.7 from the 31.2 they were polling May 1,2011 a drop of only 1.5%.

    ReplyDelete
    Replies
    1. Exactly. EKOS basically thinks nothing has changed since the election, but since then the NDP and Conservatives have lost a fair amount of support while the Liberals and the Bloc have been gaining support.

      Delete
    2. Every election is different. Ekos underestimated Tory support in May 2011, but in the 2008 election they were dead-on about Tory support. There is also no evidence that polls underestimated Tory support in the Ontario election this fall.

      Delete
    3. BCVoR, you can't just say that because the last EKOS polls were off by 5% that for every single poll EKOS publishes for the rest of eternity you have to add 5% to the Conservative result. That's wildly illogical, even if EKOS hadn't changed polling methods.

      Delete
    4. In the 2008 election ekos last, best poll on Oct 13 had the CPC at 34.8% ... only 2.8 below what the CPC "polled" in the election the next day. The correct answer in 2008 (in the back of the book provided by the voters of Canada) was 37.6

      In 27 polls published by EKOS in the 2 month prior to the 2008 election they had the CPC at 38% exactly 2 times. They had them over the 38% exactly zero times and at 34 or lower 8 times.

      It doesn't take a phd in Stats to see that EKOS has been systematically wrong in their poll results since at least 2008. For the whole 2012 election they were of by 5%.

      Every once in a while their poll would have to show the CPC at 42+ to have anywhere to a normal error distribution. Never have and I believe never will.

      The absolute only benefit is to an EKOS poll is to compare the results of the EKOS polls and then overlay on the the known results. That means adding 5% to the CPC every time.... you might make an arguement for only adding 3 or 4 % but else you should not waste your time looking at an EKOS poll.

      Delete
    5. Did you ignore the part where EKOS changed methodology? You are coming pretty close to comparing apples to oranges.

      Delete
  6. EKOS used a hybrid online/telephone panel for this poll, unlike the IVR-only polls they did in April-May 2011. Though they are from the same firm, they are not the same methodology so these kinds of comparisons are inaccurate (and they are simply misleading to begin with).

    ReplyDelete
    Replies
    1. Did the sampling methodology change?

      Did EKOS identify their problem in previous polls?

      I would not presume to say that it is biased in questions, but for them to be so consistently wrong in exactly the same direction would suggest a deeper problem in their methodology:

      Filtering valid acceptable responses;
      analysis of the results or applying the standardization of the sample to the the universe.

      Whatever their problem you do not give a consistently failing researcher credibility on their word that they have fixed the problem.

      Simply changing the interview contact process does not seem to be a panacea to all their previously inaccurate results.

      Delete
    2. You are Sir Éric, Bane of Those Who Misuse Polling Data for Partisan Purposes!

      Or, at least if you were knighted, that's the name I'd pick. You're exactly right, though; change in methodology or not, what BC VoR does is simply misleading.

      Delete
    3. LOL BC VoR's sad partisan attempt at excusing the fact he does not like.

      Delete
  7. Yes, BC Voice, the sampling methodology did change. They now use an online panel, where before they did not. That's a difference in sampling methodology.

    ReplyDelete
  8. BC Voice of Reason is just trying to shoe horn the results to fit his comfort zone. He has trouble with polls that do not help Dear Leader. It is typical of his posts on other boards as well.

    ReplyDelete
  9. I don't want a flame war so I'm just going to shut down this line of discussion.

    ReplyDelete
  10. 11 mars 2012
    Le Bloc québécois déloge le NPD ?????

    http://tvanouvelles.ca/lcn/infos/national/archives/2012/03/20120312-044747.html

    ReplyDelete
    Replies
    1. http://threehundredeight.blogspot.ca/2012/03/conservatives-lead-by-eight-in-three.html

      Delete

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.