Jean Chretien used to say that “polls were for dogs.” Others say that the only poll that matters is the one on election day. I disagree. I think it’s obvious that although polls are supposed to given an indication of voter intention, they can also influence how people vote. People tend to rely on them, assuming they are scientific and unbiased. The media report them, often as headline stories (eg. “Tories Begin Campaign Far Ahead of Liberals;” “Tories surge to 19 Point Lead”).
Many such headlines in this election, including the two cited above, have been based on Ipsos polls. Ipsos has consistently had the Conservatives polling at over 40% (ie. within majority territory).
In past elections, some bloggers have suggested a pro-Conservative bias on the part of Ipsos in the kinds of questions posed. So I thought I would take a look to see where the Ipsos results stand relative to other pollsters, going back to the beginning of the year.
Pre-election, pollsters weren’t always polling within the exact same time periods so I’ve tried to compare Ipsos results with the polls taken with those that are proximate in time.
As you can see from the table, Ipsos draws its conclusions from consistently smaller sample sizes than the other polling companies and with larger margins of error. One would think, over time, that that margin of error might favour the Liberals, but no, Ipsos has consistently either polled the Harper’s Conservatives with higher numbers than the other polling companies, or with lower ones for the Liberals. (And when the sample size and margin of error are roughly the same as those of another polling company, eg the results of April 5-7, they still have the Liberals 5% below what Nanos polled them at.)
Which means I will no longer pay any attention to their results. But other Canadians probably will.
All of this information, by the way, and more, can be obtained from the Simon Fraser University Elections website that compiles and aggregates polling data.
|Date||Pollster||Tories||Liberals||Margin of error||Sample size|
|Mar 22-23||Ipsos||43||24||+/- 3.1%||n=1001|
Update: Mere moments after posting this blog, Canada.com reported that, based on an Ipsos poll, Canadians ranked the Conservatives and NDP as getting top marks (As and Bs) for the effectiveness of their national campaigns, compared to the Liberals. Which I frankly find unbelievable, given the extent to which Harper has been hammered in the past week for the thuggery at his rallies and the steel cocoon he built around himself. Buyer beware.
Peggy’s regular blog posts on writing and getting published will resume on May 3rd, after the Canadian election.