Thursday, April 7, 2011

Change in BC, Ontario, and Quebec - or statistical wobbling

Four new polls out today, from Angus-Reid, EKOS Research, Environics, and Nanos Research, all have slightly different stories to tell. But what we can see across three or all four of these polls is much more interesting.
At the national level, the story is one of consistency. All of these polls are well within the margins of error of each other, pegging the Conservatives at roughly 38% and the Liberals at roughly 28%. The picture is less clear for the New Democrats, though, as two of these polls put the NDP at a level of support higher than they had in 2008, and two of them put the NDP below that level.

Now, of course, these polls are not all identical in make-up. Environics (UPDATE: Full release available here.) polled over one week, while the other three poll over two or three days. Environics and Nanos use a telephone survey, EKOS uses an IVR telephone system, and Angus-Reid uses an online panel. Nanos does not prompt with party names, while the other three do.

Nevertheless, there are some consistencies across these polls.

In British Columbia, the Conservatives appear to be struggling. Earlier in the campaign they were polling at 45% or so, but all of these polls put the party at 42% or lower, with three of these polls pegging the party at 39% or lower. Who takes second spot, however, appears to be less of a consensus opinion.

In Ontario, the three polls taken over a shorter period of time and more recently show a close race between the Conservatives and the Liberals. Only Environics, which polled over a longer time, appears to still be tracking the earlier, wider gap that existed between the two parties. However, despite the close race, the Conservatives still hold the advantage. There seems to be little indication that they are going to be giving it up any time soon.

On the other hand, EKOS has the Liberals up in and around Toronto (44.1% to 35%) and Ottawa (54.4% to 27.9%). But this EKOS poll was smaller than usual, so the CMA samples are tiny. We should hear from them with a larger sample at the end of the week.

In Quebec, steady Quebec, changes are taking place. But first a quick word about EKOS finding the Bloc Québécois at only 28.3%. That is one of the lowest results that has ever been recorded for the Bloc in recent memory. The last time I saw them that low was when Strategic Counsel (no longer in operation) put the Green Party at around 26% in the province. I think it is reasonable to assume the small sample size is a factor here, but what EKOS is recording is something the other polling firms have found: the Bloc Québécois is looking soft.

Prior to the campaign the Bloc was doing very well in Quebec, polling between 38% and 43% in virtually every poll. But since the campaign has started we're seeing more 34%-36% results. It still gives them a hearty lead but it appears to be the case that the Bloc is trending downwards. I don't believe the Bloc is actually down to 28%, nor do I believe that is even possible in the short term. But I do believe that the trend EKOS is identifying is a real one.

The beneficiary of their drop appears to be the New Democrats. They are doing well in all of these polls, posting second-place finishes in two of them. Angus-Reid has them four points up since March 24 and EKOS has them up three points since March 31. Nanos has seen steady increases for the NDP for five consecutive days in Quebec. It's undeniable that the NDP is making real inroads in the province. Whether the party can convert that support to seats, however, is another matter entirely.

But nationally, the parties are stable. Neither EKOS, Nanos, or Angus-Reid show any statistically significant changes at the national level since their last polls, and Environics hasn't weighed in on the federal question for some time. And to get back to this post's title, even the shifts in the three largest provinces are still relatively small. But some minor trends can be observed.

The questions asked by each of the pollsters are as follows: "For those parties you would consider voting federally, could you please rank your top two current local preferences?" (Nanos). "If a federal election were held tomorrow, which party would you vote for?" (EKOS). "As you may know, a federal election is scheduled for May 2. If the federal election were held tomorrow, which one of the following parties would you be most likely to support?" (Angus Reid). "If a Canadian federal election were to be held today, which party's candidate would you vote for?" (Environics).


  1. So if Vancouver CMA is greater Vancouver there are 12 ridings,

    Van Centre
    Van East
    Van Quadra
    Van Kingsway
    Van South
    North Van
    West Van
    Port Moody-Port Co
    New West Port Co
    New West Burnaby
    Burnaby Douglas

    This is totally unscientific, but you have an average of 33.0% for the Conservatives when averaged over these 12 ridings. I'm assuming turnout and pop. is the same for all these ridings, which it won't be. It's just a rough calculation.

    However 33% is a whole lot lower than 42% polled by EKOS, which regularly polls Conservatives lower than other polling companies.

    I think your numbers are low for Conservatives in Vancouver ridings.

    MOE is probably big, but since these riding comprise half the population of BC, not that big.

  2. 54-27 Liberals in Ottawa seems weird when the Liberals hold only two seats in the Ottawa area. The local media isn't anticipating any big swing in seats; if there was one I would think the Tory majority notion was toast.

    NDP story in Quebec is pretty interesting.

  3. I really wish people would not focus too much on the regional numbers in a lot of the polls, the statistical curve from those very small samples really makes them highly suspect.

    Personally I am only considering regional numbers with more than 300 responses - which is +-5.54 percentage points and still a very flat curve.

    Nationally I am assigning a weight to each poll based on the sample size, a fairly straight linear inverse proportional relationship to the margin of error. The smaller the MOE, the more weight. I am also only working with the latest poll of each company and stale dating them.

    What is surprising me is how far apart the various pollsters are from each other. Nanos is on a different plant from most of the others.

  4. Bernard,

    I am treating polls the same way. Weight is partly determined by the margin of error. The other factors are age and the track record of the polling firm.

    Regional polls certainly do need to be looked at with caution, but trends across several polls are worth noting.

  5. I wonder why nanos seems to consistently show about half the support for the green party than that all the other polls, seems a bit odd since the other numbers are fairly close. Hmm well actually I guess the range for Liberals and NDP is about the same, maybe its just more noticeable since the greens have a lower number to begin with?

  6. I think a real weakness with these seat projection models is how to deal with a party suddenly becoming a factor in a region where it previously had consistently low support. Any way you slice it - if you just impose a uniform swing on the NDP vote in Quebec - you won't get any new seats - even if the vote doubles to over 20%! Similarly if 308 had been around in 2006 - and polls were suddenly showing the Tories with over 20% of the vote in Quebec after having had 8% in 2004 and not been close anywhere - I'll bet the model would have projected them to win 1 or 2 seats. They got 10.

    It remains to be seen if the NDP actually ends up with over 20% in Quebec - but if they do - i think its a virtual certainty that it will mean at least 4 or 5 seats - can't say where though - maybe Hull-Aylmer, Gatineau, Outremont, Abitibi-James Baie-Eeyou, Jonquiere-Alma and maybe another wild card

  7. You should stale date all the old polls from the same polling company.... not just give them a lower weight. In elections things change quickly, and it doesn't make sense to have polls in a prediction from weeks ago. I see you give small weights to the 2004, 2006 and 2008 elections as well... that's just odd.

  8. Looks like there is something going on in Quebec.

    Latest Forum Poll:
    Bloc: 27%
    NDP: 23%

    Things just got interesting in this election!

  9. I think NDP support is probably too evenly spread in Quebec to nett them any more than two or three seats. The less BQ MPs the better IMO.

  10. Quebecers are probably waiting until after the French debate, where if Ignatieff performs well and there is a sense he might form a government they will start to move in that direction so as not to be left out they are moving away from the Bloc, but have not yet decided who to move towards because it's not yet clear what the government will look like. We'll have a much better idea in the last week of the campaign.

  11. Why ask what your first and second choice is, what happens to the second choice answers. Do they call those undecided voters or what.

  12. Is there any reason to think that the NDP support in Quebec is widely spread?

    Maybe it is all concentrated in Montreal?

    Layton seems to be the choice in Quebec to be the PM in the coalition. To me he and the NDP seems to be all urban party... Layton wrote of the rural support with the LONG GUN support.

    The people of Montreal held their noses and re-elected the adscam tainted Liberal MPs in 2008 because to them there was no federalist option. As well they still thought they may be electing cabinet ministers. No chance to think the Liberals will form the government this year.

    OR maybe the NDP has finally made an impression on the Quebec trade unionists and will take the urban unionist vote from the BLOC.

  13. I think that NDP support in Québec is still mostly concentrated to the Montréal region and Outouais. I also think that the NDP stands to gain most in the ridings that were formerly the battlegrounds between primarily the Libs and Bloc. The Laval area, Jeanne-Le Ber, and maybe a few other surprise ridings in Montréal, with their best Conservative seat opportunity being Pontiac. We'll see though, just how diffuse the NDP is among the certain regions and particular ridings in regions that are possibly in play. Also, if NDP can achieve high numbers around 28-30%, where a true breakthrough could occur.

  14. I realize the sampling sizes regional sampling sizes are small, but as an indication of trends to looks for, I look at EKOS polls removing Vancouver, Toronto, and Montreal CMA's from the provincial figures.
    The changes between yesterday's poll, and the prior one from the 1st, aren't unexpected (to me).
    The NDP Quebec growth is in the Montreal CMA, and it primarily comes from the Conservatives -- which is fine for the Conservatives as they have by far the strongest growth in the rest of Quebec and it is coming from the Liberals.
    In Ontario, Toronto's CMA sees NDP votes switching to Liberals, but in the rest of Ontario not much is occurring (The NDP has gained, but a little bit comes from all parties).
    The big move, and maybe the singular big concern for Conservatives, is the switch in BC outside of Vancouver's CMA, where the Liberals surged 10% at the expense of primarily the Greens, but also the Conservatives. From one poll to the next, the Liberals went from the fourth place party outside of the Vancouver CMA, to a very strong second.

  15. Re the NDP in Quebec, the reason why it is believed the NDP support is fairly dispersed is that the party seems to have replaced the Liberals as the default progressive federalist party among Francophones, but not among Anglophones or Allophones. So, federalist Francophones outside of Montreal who would have voted for the Liberals in the past are now thinking of voting for the NDP. But, just as the Liberals in a lot of those ridings went down to spectacular defeat against the Bloc, so will the NDP.

    While I agree with the comments above about how we should expect some surprise NDP wins if their numbers stay this high, what I really think the NDP is doing is setting the stage for the next election. If they can make themselves look viable by placing second in the popular vote or by picking up a few unexpected seats, then there is a real potential for a takeoff after Mulcair becomes leader and after Gilles Duceppe steps down. A lot of people personally like Duceppe and are willing to vote for the Bloc as long as he's there, but they might at least consider another option if he were gone.

  16. waiting for the next election is not a good strategy for the NDP. Might be valid in Quebec but will be really difficult in RoC.

    Mr. Layton is a well respected leader that the country admires and is very comfortable with. Mr. Mulclair is a time bomb. He has attacked the press much more openly than Harper.

    The next election the NDP will concentrate on getting their new leader established.


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.