Thursday, April 9, 2015

Teddy on Alberta's strange polling

I've done a bit of digging into the data behind Alberta's polls.

Before I go further, lets re-examine the recent polls:

The most shocking was a recent poll showing the NDP in first.

29% - NDP
24% - WR
23% - PC
15% - LIB

Next is another shocker from ThinkHQ

31% - WR
26% - NDP
25% - PC
12% - LIB

Next is a poll from Insights West

31% - PC
27% - WR
22% - NDP
15% - LIB

Next is a poll from Mainstreet Technologies

30% - PC
30% - WR
22% - NDP
14% - LIB

Last is an old poll from a month ago from Environics

46% - PC
18% - LIB
17% - NDP
16% - WR

There are a few clear things to note.

First, all 3 recent polls are from smaller companies with less of a quality record of being correct.
Second, all 3 recent polls show a very strong change from polls taken since the Smith defection.
Third, all 3 recent polls, however, are similar to polls from just prior to the defection, with one simple change:

The NDP is sucking up Liberal voters.

What I find interesting when I examine the tabs of the polls that have them, is that a group of voters have just vanished.

Wildrose voters who, after Smith defected, said they were going to vote PC.

These people have not seemed to have gone back to Wildrose; though due to limited information I could be wrong. This bloc of voters appears to have just disappeared. When I add them back in, and, try to average out the polls a bit, I get a more realistic number. In addition, I see a very strong trend, Liberal voters; especially outside Calgary, are abandoning the party.

While in 2012 we saw the Liberal and NDP vote drop to 10% each, this was, in part, due to people voting for Redford to stop Smith. Going back to 2008 we see the Liberals took 26% and the NDP 9%, for a total of 35% of the vote on the "left" With that kept in mind, it is therefore not so crazy to say the NDP is possibly at 30%, and makes much more sense if the Liberals, who appear to be doing very poorly, are, actually, at 5%.


The NDP actually is doing very well.
The PC Party actually is doing poorly.
Wildrose actually is competitive.
And none of these 3 are to the extreme that more recent polls have tried to tell us.

Due to all that I feel a more "realistic" number for the parties as as follows:

38% - PC
27% - WR
23% - NDP
12% - LIB

Just to re-iterate, this is my personal prediction of where the polls stand, with an adjustment for what I feel will happen to the Liberal vote.

The good thing about this shocking poll result mentioned earlier is it will encourage more reputable companies to poll the province to try to figure out what is going on. Hopefully one of them will have that crucial tab that has always served me well in finding polling flaws - raw numbers of respondents and how they claim to have voted in the past election.

Consistently and without fail this tab has let me know one thing; who is staying home. If the raw data shows that one or another party has fewer people who claim to have voted for them, it indicates that these people have been turned off not just their party, but politics, and will be unlikely to vote. Pollsters try to weight the parties back to their last-election level, but doing this only doubles the error in their polling. When I take out that error and go with the real base numbers, I almost always get polling results that match the real results in elections where polling and the results are quite distant.

One thing to note is that this personal prediction of mine ends up being quite close to our Blunt Objects Official Projection, with the only major change being my predicted Liberal-to-NDP swing.

I can't see the Liberals holding on to any seats outside Calgary, with the possible exception of Blakeman. Meadowlark is a Raj Sherman seat, not a Liberal seat. The last time the Liberals took it without Sherman, they were far ahead of their current position in the polls. I'd be less surprised to see the NDP take 14 seats in Edmonton. which, frankly, may well happen.

In Calgary I also see some problem spots. Without Kang, who so heavily won the advance vote last time that media outlets who went to print before they were out said Wildrose had won it, McCall may be in trouble. The only two ridings that are certainties for the party are Buffalo and Mountain View. At this time I'm likely to give McCall to the Liberals on the grounds that we've still yet to see any total collapse in the Liberal vote.

What that gives us is this:


  1. Very interesting about the "disappearing" Wildrose voters. One thing I noticed from one of the polls was the high undecided bloc 25%, I suspect some of the disappeared are in this category. In Edmonton some of these voters must have turned to the NDP, if one believes the polls

    I really don't think the Liberals are at 5%. If they can pull off 10% last election that is probably their low-base. As you note two elections ago they received 26% it would be improbabble if today they are retaining less than 20% of that vote. Having said that I think you are correct about Meadowlark and although at this point I believe Blakeman is the favourite to retain her seat, the Edmonton-NDP wave certainly puts Edmonton-Centre in contention. Although Blakeman is both the Liberal candidate, Green Party candidate and Alberta Party candidate which probably plays to her advantage at least in theory. I don't think there is any evidence to suggest Green and Alberta Party voters will vote for Blakeman en masse but, even an extra 1 or 2 per cent could be critical in a tight race with the NDP.

    Over all I tend to believe these polls and so, I think your projection of 54 seats for the PCs overly-generous. However, I am far less certain that the polls will stay as they are between now and May 5th. I would expect a good degree of fluctuation in the polls and so at this stage I think the outcome of the election is unknowable.

    1. My own theory on Blakeman is that her decision to gain the other endorsements will only benefit her peripherally - those Greens and Alberta Party voters are not going to worry about voting for Blakeman if they don't see a right-wing candidate as a contender, they'll vote for whichever centre-left party they want. If the NDP become the main contenders, I feel Blakeman is more likely than not to lose her seat.

    2. I agree that Blakeman is only likely to benefit marginally of peripherally, but, I believe this will boost her vote total slightly, the question is whether such a benefit will be enough to stave off an NDP wave? I agree with your assessment of the NDP although Blakeman does have a strong connection and following in her riding.

  2. You think the NDP will win McClung and Rutherford but not Castle Downs or Ellerslie? Is Lukczuk too popular?

    What about the NDP in Calagry-Fort. The candidate, Joe Cici represented the area as an alderman for 15 years.

    What about Alberta Party leader Greg Clark's prospects in Calgary-Elbow? He nearly won the by-election last year

    1. Calgary-Elbow is a tony-upper class neighbourhood. With the polls the way they are I wouldn't write this is a safe PC seat but, I think they still must be favoured. Greg Clark lost the by-election by 800 votes and 6%. Alison Redford won the riding by 6,000 votes and 30%. Partly the difference is a reflection of dissatisfaction with the PC Government last Autumn but, part of that difference is the difference between by-elections and general elections as reflected in turnout.

      The one caveat I will make is that if it appears a Wildrose Government is on the horizon Calgary-Elbow will vote Wildrose. When Albertans change Government they go whole hog and the people living in million dollar plus houses in Calgary-Elbow will not want to miss out!

      Calgary Fort is a different story, the NDP is only polling at about 16% in Calgary, yes, that support is more concentrated in some areas than others but, with the Wildrose and PCs polling roughly 10 points higher and the Liberals holding their vote Ceci is in a tight three way race in which he is frankly the underdog. If NDP fortunes continue to rise he could pull off an upset but, at this point Calgary Fort looks likely to go either PC or Wildrose.

      Castledowns and Ellerslie probably have a better chance of going NDP than Calgary-Fort but, it depends on certain factors such as turnout. We know older people are more likely to vote and they generally favour the PCs. That's not to say The PCs will win but, everyone who makes a seat projection will have their own methodology, you may not like Kyle and Teddy's projection but, I think they probably have a method to the madness.

  3. Follow me on twitter to see my updates