Election Going From “Unnecessary” To Wild And Wacky

I stumbled upon this article from the punditsguide.ca blog about how various websites that predict seat counts fare when compared to actual election outcomes. One of the big points of the post is how no polling site really gets it right so maybe they shouldn’t be trusted. Fair enough. However, of the sites it surveys, Ekos does a pretty okay job. Have a look at the numbers the punditsguide sites….

2008 election actual: Cons 143, Lib 77, BQ 49, NDP 37, Other 2, Green 0
Ekos predicted: Cons 136, Lib 84, BQ 51, NDP 35, Other 2, Green 0

As you can see, Ekos wasn’t that far off. The actual count varied from their prediction only by 18 seats all together.

So, just for shits and giggles (and accepting that I really shouldn’t be concluding that Ekos is especially competent based off one data point), with a week left of campaigning, what are their seat projections for the 2011 election? Well, as of April 26, this is how they see things….

(By the way, that chart is plucked from their seat projection pdf which you can download here.)

As you can see, Ekos is presently predicting the NDP for the official opposition. And by a hefty margin, to boot. And with the Libs holding enough seats to give the two parties majority control of the House. Yowza. I have never seen the federal NDP polling so well. This is Prime Minister Layton territory. I wonder how deep into Ignatieff’s craw that would stick?

Of course, we must remember support for progressive parties always collapses on election day. So goes the conventional wisdom. And at least some of that collapse can be traced to the fact that the people who may say they’re going to vote progressive end up just not voting — whether because of apathy or feelings of disenfranchisement or whathaveyou. Low voter turnouts hurt parties like the NDP the hardest.

So I guess we should ignore these anomalous polling numbers for the NDP seeing as, in this “Time To Lead” feature from the Globe and Mail website that I’ve recorded in a screen cap at right, we can expect to see voter turnout plummet to a new low this election. Sigh….

Wait, what’s that headline just above the “Time To Lead” feature? An unexpected record high turnout at the advance polls?

Oh. That is surprising. Very surprising indeed….

Author: Paul Dechene

Paul Dechene is 5'10'' tall and he was born in a place. He's not there now. He's sitting in front of his computer writing his bio for this blog. He has a song stuck in his head. It's "Girl From Ipanema", thanks for asking. You can follow Paul on Twitter at @pauldechene and get live updates during city council meetings and other city events at @PDcityhall.

15 thoughts on “Election Going From “Unnecessary” To Wild And Wacky”

  1. It’s alright to look at the macro-level seat projections … so long as you understand all the methodological caveats.

    What’s really unsafe is extrapolating those down to the seat-by-seat level 30 days out from E-day, and then using them in a “strategic voting” site to generate voting recommendations.

  2. Holy MoFo. While I think 97 doesn’t add up for the NDP, anything above 60 would be a-ma-zing. The NDP WILL NOT win 51 seats in Quebec. More like 10-15. And they probably won’t win zero here. More like 1-2. I was a little worried Jack may have been overexposing himself last night during hockey: I don’t think that’s a good time to bombard viewers…extremely stressful and all.

  3. Exactly. I would be very surprised if at least one seat in SK didn’t go orange.

    But yeah, who’d have thought? A week ago, if you said “PM Layton”, you’d get sniggered out of the place. Now…it’s not so funny anymore.

  4. I notice that Ekos says that the NDP will get 0 seats in SK. Other polls i Have seen put the NDP as winning in my riding? Certainly walking around town it would seem that the NDP candidate in my area has a great deal of support?

  5. When I was working for a pollster, they used to do their polling on the census division level, using that to predict the ridings, then aggregate the data to make predictions on the provinces and country. I would strongly expect that is how it is still being done, as opposed to surveying the country as the primary unit.

    (If you understood that, I feel sorry for you.)

  6. Greg Harris (of Harris-Decima Research) says there are huge problems with polling these days because:

    a) Respondents no longer answer their phone when a weird number calls
    b) People don’t want to be bothered by taking part in a poll or have no time
    c) Cell phones now outnumber landlines.

    He said they used to get something like 70-85% return on their calls, now they get something like 15%. Statistically, it is supposedly still possible to get a good sample, it just takes a lot more work.

  7. Odd that none of those NDP seats are in Saskatchewan. Seems to me that Brian Sklar and Fred Clipsham have a pretty good shot…

  8. “Unexpected record high turnout at the advance polls” is good news no matter how you slice it. People voting = good. But for proggys that may turn into good news as well.

  9. @ Moon Daddy: I hear ya, but that’s what they said in 2004. “Oh, John Kerry is 4 pts down from Bush in national polls BUT the wild card is cell phones!! WHO OWNS CELL PHONES?!? THE YOUNG!?! WHO VOTES DEMOCRATIC? THE YOUNG!!” On and on… Anyway, they predicted Kerry would make up 5% based on cell phone-voting, non-poll-reaching young voters.

    Never happened.

    If Nanos and company can draw a conclusion for 20+million potential voters based on a 989 sample size, I’m pretty sure cell phones they can do w/o.

    BTW, I noticed Ekos polls 2X as many people and have consistently lower totals for Cons, so that’s a good thing.

  10. I guess we’ll see, Talbot. But fuck everything about American elections and their hackable voting machines.

  11. EKOS does 50% cell, 50% land lines. It’s in their “Methodologies” section when you look at their reports online (PDF for latest projection)I’m tired of the “Nobody calls cell phones” bit, it’s not applicable anymore. And I’ve made it my job to answer polls, even if I’m going out the door. Gotta get the 25-29 year old female demographic some representation!

    I would regard riding projections with suspicion, though. But you must admit, these are exciting times for the NDP.

  12. I still suppose that Alberta skews things, too, tho, screw us for having no NDP seats and Alberta for having infinitely more than none.

Comments are closed.