Home » Election Polling Isn’t Supposed to be Helpful

with 3 comments

Kennedy Stewart is an SFU Public Policy professor: I’d classify him as a sociologist of politics. He has a model that takes province-wide poll numbers and converts them into seats. The Tyee has an interesting little interview with him where he explains where polls come from:

The reason why pollsters release results during elections is to demonstrate their accuracy so they can sell polling services to private/government firms after the election. Elections are a perfect way to demonstrate polling accuracy as projections can be checked against actual Election Day results.

Economic theory says that should cause polls to converge because it’s safer to compete on cost or service. Of course a market this small is bound to be full of fail, so instead we have polling companies competing on methodology. The punchline to this is that the Election Prediction Project has horrible methodology but generally does a pretty accurate job.

Written by Jared

May 6th, 2009 at 10:27 am

Posted in Uncategorized

Tagged with ,

3 Responses to 'Election Polling Isn’t Supposed to be Helpful'

Subscribe to comments with RSS or TrackBack to 'Election Polling Isn’t Supposed to be Helpful'.

  1. How is the STV referendum shaping up?

    You’ve probably seen this Google Maps map from the CBC where you can turn the proposed STV riding boundaries and the current 2009 provincial riding boundaries on and off.

    Don

    6 May 09 at 10:34 am

  2. All the polling done for STV that I’m aware of is push-polling, so the results aren’t useful.

    Jared

    6 May 09 at 10:49 am

  3. All the polling done for STV that I’m aware of is push-polling, so the results aren’t useful.

    Well, I want to know: would you be more likely or less likely to vote for STV if you knew that it had fathered an illegitimate black child?

    Don

    6 May 09 at 1:40 pm

Leave a Reply

Notify me of followup comments via e-mail. You can also subscribe without commenting.