Posted Sep. 1/05
Previous: Behind the scenes
DEAD HEAT.” “Down to the wire.” “HARPER WIDENS SEAT LEAD.”
Prompted by national opinion polls, these were some of the headlines from national media outlets in the days before last year’s federal election.
Some polls triggered news stories that predicted a razor-thin Conservative victory over the Liberals, making Conservative Leader Stephen Harper the country’s next prime minister. Buoyed by such reports, a jubilant Harper boasted mid-campaign that his party was preparing to take over the reigns of power from Prime Minister Paul Martin.
But these heady predictions and declarations did not come to pass, which begs the question: What went wrong?
According to Darrell Bricker, PhD/89, president of Ipsos Public Affairs in North America, Canadian voters simply changed their mind at the last minute.
“It doesn’t usually happen,” says Bricker from the polling firm’s Toronto offices. “Usually, you can pin it a week out and get it within the margin of error. But there was enough of a change at the last minute among about six percent of the vote in Ontario to shift the entire election.”
It was also an issue of timing. The election took place on a Monday, so in order for pollsters to get their final results published in the weekend papers, they needed to submit four or five days before voting day, says Bricker, adding that by the time Canadians cast their votes, they were reading about poll results that were nearly a week old.
“The polling wasn’t inaccurate, the election moved on,” says Bricker. “And you can end up with some critical movement in the last minute that moved a lot of seats, particularly in a close election. We haven’t had a close election like this since 1979.”
- Christopher Waddell, Carleton’s Carty Chair in Business and Financial Journalism
Indeed, good public opinion polling will give you a snapshot of what people were thinking at the time the poll was done, says Christopher Waddell, Carleton’s Carty Chair in Business and Financial Journalism and a former Ottawa bureau chief with the Globe and Mail.
But polls can never tell you anything about what people will think in the future about a particular issue, adds Waddell. The problem is, this doesn’t stop the media from using polls as a device to try to predict what people are going to think a few days later, he says.
This means the public needs to be on guard when interpreting polls, says Waddell, adding that while the sample size and questions of various polls can be different, the media tends to equate all public opinion polls as the same.
“Let’s say you went out and asked people, ‘If a federal election were held today, which party would you vote for?’ If you ask that question at the start of a poll or at the end, you get different answers. There’s a debate within the polling community about the best thing to do.”
For Bricker, the best way to tackle this is to always ask the main question at the start of the survey.
“That’s my view, and we’ve been doing it for years. There is evidence showing this yields accurate results.”
Another challenge for pollsters is the public’s intolerance to telemarketers, which have brought response rates down to about 20 percent from 50 percent, says Waddell.
“Is there a segment of society that is more likely to have ways of avoiding telemarketers (such as answering machines or cell phones)? Could it be that (pollsters) are missing a whole demographic of society?”
Bricker says this isn’t an issue. He points to the U.S., where there used to be growing concern over missing a large group of young people who only owned unlisted cell phone numbers.
“But if you look at the American polls, they’re all spot on,” he says. “And it has nothing to do with response rates.”