Thursday, January 26, 2012

Two Alberta polls, two outcomes

With the Alberta legislature set to get back to work on February 7 and the next election campaign likely to begin at the end of March, two polls were released this week for the province. One was by Forum Research and the other was by Léger Marketing. The two were radically different from one another.
Forum indicated that Progressive Conservative support stood at 38%, unchanged from their last poll in December. The Wildrose Party had gained six points to reach 29%, while the Liberals were up two to 14% and the NDP was steady at 13%.

But Léger Marketing found that Progressive Conservative support was at 53%, with the Wildrose Party at 16%, the NDP at 13%, and the Liberals at 11%.

Let's start with where these two polls agree. Both seem to have pegged NDP and Liberal support at about the same level, and even across the regional breakdowns the polls line up quite well. So, we can say with some confidence where the Liberals and the New Democrats stand.

But this race is going to be between the Tories and Wildrose, and on this score the two polls differ greatly. Léger gives the PCs a 37-point lead, while Forum gives them a nine point lead. That is the kind of difference that the margin of error cannot explain away.

Why are these polls so different? Forum uses the IVR method to conduct its polls while Léger uses an online panel. EDIT: Léger uses an online panel for its Quebec and federal polls. In this case, Léger contacted Albertans by telephone. While that might be one reason for the difference, this kind of disparity should not happen. Both were taken at about the same time, with Léger in the field between January 13-18 and Forum in the field on January 17.

But the difference between a poll taken over six days and a poll taken over one day can be huge. A poll taken over a week will be less influenced by day-to-day events, whereas a poll taken on one day might be overly influenced by whatever was in the news that day or the day before. However, there does not appear to have been a major story that broke at around that time, so it is difficult to say what could have provoked such a backlash against the Tories to the benefit of Wildrose. There was some talk of dodgy political financing, but Léger should have captured that as well. If Léger had a day-by-day breakdown, we'd have a better idea of what happened.

There is a small difference in how Forum and Léger defines its regions, but aside from Edmonton (by my calculations, Forum's Edmonton numbers would be 38-20-17-17 if they used Léger's definition of the city) it does not change much.

Since Forum has shown little real change since their last poll from December while Léger Marketing seems to line-up with some of the other polls that were taken at the end of last year, we might have to conclude that there is a methodological reason for this difference. That does not help us determine which of these two polls is closer to the truth, unfortunately.

The polls do agree on a few things. Both Léger and Forum indicate that PC support is generally uniform across the two major cities and the rest of the province, while Wildrose is strongest in Calgary. They both indicate that the NDP is running second in Edmonton while the Liberals are also doing best in that city. Their support in Calgary, which was relatively high in the last election, appears to have bottomed out.

But, again, the problem is the margin between the Tories and Wildrose. Forum sees a very close race in Calgary and a close one as well in the rest of the province. Léger has the Progressive Conservatives romping to victory across Alberta.

This is reflected in the seat projections for these two polls. The projection model for Alberta is completed, and incorporates a few of the lessons of the 2011 provincial campaigns. The projection model is regional, breaking the province down into Edmonton, Calgary, and the rest of Alberta, and has a few other tweaks that differentiate it from the models used in past elections.  I'll go into more detail when the projection model is fully launched.

I will start making projections for the Alberta election once the legislature returns. I am still working on a vote projection model that should bridge the gap between what the polls say and what the voters do.

Plugging these two polls into the model individually gives very different results. With the Forum poll, Wildrose would make a big breakthrough and form the largest opposition since 1997. With the Léger poll, the opposition ranks are reduced and the NDP becomes the second largest party in the legislature.

More specifically, with the Forum poll the Progressive Conservatives would win 65 seats, 18 of them in Calgary, 24 in Edmonton, and 23 in the rest of the province. The Wildrose Party would win 17 seats, half of them in Calgary and half in the rest of the province. They would be shut out of Edmonton. The New Democrats win four seats, all in the provincial capital, while the Liberals win only one.

With Léger's numbers, the Progressive Conservatives win the largest majority government in the province's history, with 81 seats. The New Democrats win four seats, all in Edmonton, while Wildrose wins only two (one in Calgary, one outside the two largest cities). The Liberals don't win a single seat.

As you can see, these are two very different scenarios. In the first, Wildrose is able to put up enough of a fight to win a good chunk of the province's seats, but they are still too thin to put up a real challenge to the Tories. In the second scenario, the Tories take advantage of a very divided electorate to win a huge majority.

Reconciling these two polls may be impossible. We'll have to wait for more data from other sources before we can say with any confidence what is going on in Alberta. What we do know is that the Tories are still very well placed to win yet another election in the province.

The Forum poll had an interesting breakdown of the approval ratings of the leaders. Both Alison Redford and Danielle Smith had net positive ratings, but whereas Smith had relatively uniform approval among both men and women, Redford has terrific approval ratings among women but has a negative approval rating among men. Gender does not appear to be an issue for Smith, but it might be for Redford.

Nevertheless, according to Léger Redford is still the favourite person to be premier at 37%. Smith is at 16% while Brian Mason of the NDP comes in third at 8%.

As we approach the election, more and more attention will be turned towards Alberta. If a scenario like Forum envisages plays out, it will be an interesting contest. Redford's Tories look pretty safe, so the real question will be how Smith performs during the campaign and whether the Liberals can survive.

24 comments:

  1. Its worth noting that just a few months ago the Alberta Liberals were all exited about how they had over 8,000 people vote in their leadership contest which was open to "supporters" and not just "members". Sherman won - and now he seems to be leading them on a one way ride to oblivion.

    ReplyDelete
  2. There are only 19 seats in Edmonton and this projection has 29 seats (24 by the PCs alone) being won there. Maybe you mean the PCs will take 14 in Edmonton?

    ReplyDelete
  3. Brad, it depends on how you define Edmonton. Most pollsters (including Leger) use the CMA, which has part or all of 29 seats in total.

    ReplyDelete
  4. And this model uses the new 87-seat electoral map for Alberta.

    ReplyDelete
  5. A couple of days ago several of the residents in this town and surrounding district received a call from a number in Atlantic City, N.J. The question asked was, please take 15 seconds to answer this question, who will you vote for in the coming provincial election, press 1 for WRA. Have no idea who the other numbers were for as I pressed #1 and got this reply, thank you for voting. Caller ID said New Jersey, and the # is 609-625-4399.
    No one seems to know who is behind this poll.

    ReplyDelete
  6. These polls are not valuable until an election begins.

    Albertans are very apolitical. Only 40% of Albertans vote.

    They say they'll vote PC because they have fond memories of good ole Ralph Klein.

    Just wait....

    ReplyDelete
  7. The Liberals probably won't be shut out, and the NDP is unlikely to form Official Opposition due to the WRA's strength. The PCs are probably going to increase their share of seats and the WRA will not get as many as 17 but not as few as 2 seats.

    Maybe somewhere in the middle like 75 PCs, 8 WRA, 2 each for Liberal and NDP.

    ReplyDelete
  8. Anon 10:41,

    I would disagree. The ALP could very easily be shut out - they are, after all, losing their most popular incumbents in Edmonton, and it's easy to be squeezed out of Calgary when you're caught between a PC-WRA fight. It's more likely than not, especially at such low levels.

    ReplyDelete
  9. The two polls highlight the general weakness of polling in Canada. I see few companies explaining why their polls are more often wrong than right.

    I see little evidence of methodological rigour in the work done. I see no discussion of where the weaknesses are in their systems and how much of impact this has on the results.

    In this case one or both of the companies are completely off of the mark. Even though the results for the NDP and Liberals the same, you can not accept some data from each poll as right.

    I would like to see both companies explain why their results are so far apart and admit what their weaknesses are. I know I am dreaming because I know no one will hold them responsible for data.

    ReplyDelete
  10. Something not mentioned is how many mla's are not running again, due to retirement or losing the nomination.
    Our mla lost to a newcomer.
    In addition, Redford will only be on one ballot, so all those non PCs who voted for her will have to find someone else, and it will not be a PC. Lots of ballots will have new, unfamiliar names on them. And as most ballots are marked either with the first, middle or last candidate listed, it will depend where your party comes on the ballot.
    It will be an interesting election with interesting results.

    ReplyDelete
  11. I don't think the WRA and the NDP's fortunes are really tied to each other in any meaningful way. The NDP will do well in Edmonton and the WRA will do well in Calgary and elsewhere, both for some definition of 'well'. You can see this in both polls, though obviously moreso in the Leger one. They just don't project strength in the same areas, and in the system we have that means they don't interfere with each other much.

    The Liberals stand to lose the most from the rising fortunes of the other two parties, though. For reference, the previous elections' results of 9 Lib seats were obtained with 26% of the popular vote.

    I suspect there'll be more of a fight between the WRA and the NDP for opposition than most people think, but there's no real hard data behind that. I'd be a bit shocked if the Libs pull off any more than 1 or 2 seats (and I actually hope one of them is mine).

    ReplyDelete
  12. I'd bet on two seats for the Liberals myself: Hugh MacDonald and Laurie Blakeman. Hopefully one of them can rebuild the party.

    ReplyDelete
    Replies
    1. Hugh MacDonald isn't running again. If the Liberals are reduced to two seats, I expect his will not be one of them.

      Delete
    2. Oh he's not? That's unfortunate. And yah, I agree.

      I have p big concerns with Raj Sherman lol.

      Delete
  13. A friend of mine had a thought re: the different results. The Forum poll used an IVR, whereas Leger used real people asking the questions over the phone (Eric, you said they used an online panel but their report, which you linked to, says the people were contacted by telephone).

    This matters because I think in general, most average people are more likely to hang up on a recording than on a real person. Whereas Wildrose supporters, being generally more vigourous in their support and political action, would be very likely to want to take part in a poll. Thus Wildrose voters would be over-represented in the total, and your average Albertan, who is not that politically active but who generally votes PC because that's what they always do/have done, will be under-represented.

    No way to prove it, I guess, but it's a thought.

    ReplyDelete
  14. Anonymous 1:28, thanks for the correction, I've fixed the post. Léger uses an online panel for the vast majority of its polls, so I just assumed. Léger does have an Alberta office, and apparently they use the telephone. I'll be more careful to double-check the methodological statement next time, as some other polling firms use different methods depending on the survey.

    ReplyDelete
  15. I agree with Anonymous (1:28). Although with low voter turnout, if Wildrose supporters are more vigourous, they might have a higher turnout... leading to an actual result between the two polls.

    ReplyDelete
  16. Hi Eric. Two things to watch in AB polling:
    1-Gender distribution. Forum poll was 57% female and there is a huge gender gap vis a vis Wildrose and PC.
    2-Wording used on the Poll to describe the Alberta PCs. Polls that call them the Alberta Conservatives or the Conservative Party result in higher number because of confusion with the more popular Conservative Party of Canada.

    ReplyDelete
  17. Something else that may sway voters is if the survey lists the party leaders' names along the with party name in asking for the vote preference. For example, the Leger poll did this, asking voters if they wanted to vote for "Alison Redford's PCs" or "Danielle Smith's Wildrose". Because of Redford's general popularity, this could lead some people to pick the PCs if they like Redford, even if they might be undecided or pick another party if they just heard the party name.

    ReplyDelete
  18. Interested Albertan13 February, 2012 15:08

    Hi Eric,

    Any idea why the huge variance between the Abingdon Research and the Return on Insight (ROI) poll? Do you think this has something to do with methodology, considering the former was considered a "push" poll?

    ReplyDelete
  19. I'm afraid I don't know much about the Abingdon poll aside from a mention in a Calgary Sun column.

    ReplyDelete
  20. These polls seem politicized. The following is most suspicious:

    1. The Abington Research, Forum and Leger and ROI polls are fairly consistent on the Liberals and NDP, but vary wildly on the PC's and WRP.

    2. Both the PC's and WRP have significant money; roughly $3-million and $2-million, respectively. Given this, does it not seem plausible they have commishioned the polls? If not them, then a supporter with deep pockets and vested interest in the outcome (Big oil, or labor unions come to mind).

    Given this, it would be a huge benefit if there was a clear explanation of the methodologies associated with each poll, explaining how the firms arrived at these numbers and who commissioned the polls, themselves.

    ReplyDelete
  21. -Apologies, I did not sign my last comment.

    E.M. Boisvert

    ReplyDelete
  22. No, it is not plausible, except for Abingdon which I think was specifically commissioned by and presented to Wildrose (it was not publicly released, only mentioned in a column).

    Leger is QMI's official pollster, and their poll was commissioned by the Sun. ROI was commissioned by the CBC. Forum's poll was done by the National Post.

    Polls commissioned by parties or political organizations are not reported by the media as independent polls.

    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.