Posted on 06/29/2004 9:50:23 PM PDT by conservative in nyc
|
Tue Jun 29, 5:25 PM ET
|
|
STEPHANIE LEVITZ
OTTAWA (CP) - No mea culpas were on offer from the nation's pollsters Tuesday after election results threw five weeks of polling predictions in their faces.
Instead, those charged with taking the pulse of the nation stood by five weeks of polling that suggested Monday's vote would be a cliff-hanger. "People walking around out there generally saying 'how can the pollsters screw up' well, the pollsters didn't screw up," said Darryl Bricker, president of Ipsos-Reid.
"If you would have held the election last week, that's what would have happened."
In the week leading up to the vote, all of the major polling companies had predicted a vote breakdown that suggested the Liberals would capture about 32 per cent of popular support, followed by the Conservatives at 31, the NDP 19 and the Bloc Quebecois around 12 per cent.
The close of polls on Monday told a different story, with only Bloc support staying constant.
According to the results from Elections Canada, the Liberals won 36.7 per cent of the popular vote, the Conservatives 29.6, the NDP 15.7 and the Bloc 12.4.
"The major errors, if you want to call them that, were first of all assuming that election day was going to resemble last week which it didn't," said Frank Graves, president of Ekos.
The shift from the razor-thin edge dividing the two leading parties to the seven-point gap that ultimately gave the Liberals a minority government was attributed Tuesday to the last four days of the campaign.
With media outlets forbidden from publishing new polls on election day, pollsters had stopped surveying Canadians by June 24 to allow final polls into the weekend newspapers.
And then, people changed their minds.
As simple as it may sound, that's how pollsters explained away the discrepancy between their research and the results.
"I think Ontario was really the key," said Donna Dasko, vice-president of Environics.
"Voters hadn't settled into their vote yet and were still thinking about their decision."
In that province, the Liberals walked away with 44 per cent support, where they had been predicted to win around 40 per cent.
Pollsters pinned the swing on the success of attack advertising.
"When you have an electorate that is uncertain, negative attack ads work," said Tamara Gottlieb, a pollster with Compas.
That company did an election day poll suggesting that one in four voters had made up their minds in the last 24 hours of the campaign.
Gottlieb said that typically change-adverse Ontario voters bombarded with attack ads in the last weekend of the election forced their pencils to place the X in the safer Liberal spot on Monday.
But the shift may also be the result of undecided voters seeing the narrowing horse-race between leaders and voting strategically to keep a Conservative government at bay.
"If (voters) had thought, for example, that the Liberals were going to be in a comfortable majority position I can assure you they were going to stay with the NDP," said Graves. "It wasn't based on affection for the Liberals and Paul Martin (news - web sites), it was based on concern for where Stephen Harper and the Conservatives were going."
But with the NDP's vote share still higher than it was in the last election, something else could have been at play in the last days of the campaign.
"It may be that there was a systematic understatement of the Liberals," said Richard Johnston, a political science professor at the University of British Columbia who logged each poll released throughout the elections.
He found that the higher the percentage of undecided voters in a poll, the higher percentage was of Tory and NDP supporters.
"That was sort of consistent with the psychology of the campaign," he said. "People wanted to punish the Liberals but weren't sure they were prepared to punish themselves in the process."
Are the polls in our Presidential race also wrong? Will we see the same phenomenon in our election as in Canada -- the undecideds breaking to the "devil" they know (President Bush) instead of the "devil" they don't know (Kerry)?
Or are polls now biased due to technology --- those who have caller ID don't pick up. Can people who only have cell phones be reached by pollsters -- their phone numbers are unlisted? Are we going back to the early days of Gallup where all the polls showed Roosevelt will lose because more Republicans than Democrats owned phones?
I guess what I'm asking the more poll-oriented FReepers is whether we can gleam anything from the Canadian poll experience. Your thoughts would be appreciated.
PING
I would say yes, I've heard of people getting telemarketers and other riff-raf calls on their cell phones, pollsters probably no exception here.
Polls are measures of people's thoughts at a specific moment in time. They will fluctuate with events and even during seemingly torpid, uneventful times. There is no way that you can rely upon polls/surveys as your sole statistical source, if you're intent on discerning the will of voters. It's simply impossible.
Not only can polls vary wildly, but the questioning used and the manner in which data is double-checked (when it can be) makes a difference. For example - take a pet peeve of mine. JD Power comes out with their dependability study today, which has Mercury with almost the exact same number of problems per 100 vehicles as Toyota and Honda, but Ford is just below average. Ok, so that means that the Fords without Mercury counterparts are bad, right? Wrong! The Full sized trucks are coming in as industry leaders. So how can a Ford be less reliable than a Mercury when they're identical mechanicals, built on the same line by the same people? Someone didn't double check what people were telling them... same thing happens in political polls. I hate the things. Often I would like to qualify an answer, but they don't check that....
When I get polled, I hang up.
Disclaimer: Opinions posted on Free Republic are those of the individual posters and do not necessarily represent the opinion of Free Republic or its management. All materials posted herein are protected by copyright law and the exemption for fair use of copyrighted works.