Friday, April 22, 2011

I call BS on Fair Vote Canada

Apparently, Fair Vote Canada, which brought you this doozy of a PR projection whereby you can give the Bloc Quebecois, which only runs a maximum of 75 candidates, a 100-seat majority, has apparently come out a projection on that latest Ipsos Reid poll which gives the Conservatives 201 seats (!!!), the Liberals 53, the NDP 48, and the Bloc 4 seats.

Hmm. According to Canada.com, which Fair Vote gave/made these projections for, FV uses a "... a matrix that predicts how a party's supporters in a previous election will vote in the next election."

The "method," apparently ".... assumes a party polling above its previous popular vote will, on balance, keep all of its voters and capture some from other parties. Conversely, a party polling below last election's popular vote will be losing some of its voters to other parties."

Erm, in other words, they used a basic swing projection. Except I have no idea how they got to 201 for the Conservatives... or do I?




I call such bullshit on Fair Vote Canada, with their crappy calculator and silly notions of what constitutes accuracy and a proper projection matrix.


Using Too Close to Call's well-thought-out projection system, we get the following:

177 Conservatives, 49 New Democrats, 48 Liberals, 34 Bloc Quebecois

That's horrible but still a hell of a lot more infinitely plausible.

5 comments:

  1. Ipsos-Reid has generally given the Conservatives huge leads throughout the campaign, has pegged them at 42%-43%...

    I read somewhere that they use a variation on that "likely voter" measurement as is now standard in the USA. In this case, it's based on the hardness of the Conservative vote compared with others. That is, more Conservative voters say they're sticking for sure with their choice than the voters of other parties are saying the're doing so, and they're using this in the projection.

    The problem with this methodology is that there are many Anyone but Harper people out there who aren't sure whether the Liberals are the best bet to block the Conservative majority or the NDP is and so there's movement between these parties, and also between the BQ, NDP and Liberals in Quebec.

    Ipsos-Reid takes this and responds by adding weight to the Conservative votes thus boosting the percentage it gets in their polls.

    ReplyDelete
  2. So all these reports that claim a huge breakthrough for Harper, a great big lead opened up, this is false. The Ipsos-Reid polls always have been this way during the campaign. The only change is between the NDP and Liberals as compared with previous polls. The trends shown are identical to those shown by Nanos or by any of the others. Now, will their methodology be vindicated? Will Canada get a nasty surprise on election day? I certainly hope not!

    ReplyDelete
  3. The Nortel disabled employees need the young Liberals to stop a Conservative majority. Watch their story at

    http://www.youtube.com/watch?v=Ak8NEqo2po8

    There are a series of YouTubes on Prime Minister Harper Allows the Powerful to Abuse Disabled...Who... Lawyers, John Baird, Banks

    ReplyDelete
  4. You have been misinformed.

    Fair Vote Canada has made no election predictions.

    None.

    Our calculator is an educational tool intended to show how our voting system distorts the results of our elections.

    It is a calculator, not a predictor.

    You can put any numbers you want into the calculator. If the numbers are sufficiently stupid, it will generate stupid output.

    Somebody else put questionable poll results into our calculator, and passed the results off as a prediction by Fair Vote Canada.

    Fair Vote Canada has made no election predictions.

    None.

    Wayne Smith
    Executive Director
    Fair Vote Canada

    ReplyDelete
  5. Wayne Smith,

    You should have taken up that with the fellows who attributed it to you, then. This is not me making the claims that you're the ones who did it - I got it off the Canada.com story.

    ReplyDelete