s

Tuesday, May 23, 2006

Poll Ridiculousness, Part XVI

Lots of fuss over the latest poll showing surging support (40% is considered a surge? And we still can't agree on the need for Proportional Representation? -ed.) for Harper's Tories. As this fits in so harmoniously with the meme of the moment, the opposition parties must have cause for great alarm and the Conservatives for jubilation.

Well - here are the results of another poll from Ekos. The question: "if a federal election were held tomorrow, which party would you vote for?" The response:
Liberals - 40.2%
CPC - 26.5%
NDP - 18.7%
BQ - 11.0%
Green - 3.7%

Hold on a minute, can that be right?? Indeed it is. Admittedly, this particular survey was conducted on February 14th, 2005 - but that only furthers the main point.

Which is this: Here's a relatively recent snapshot of the populace's opinion a week before the release of a popular budget, a few months after an election and with another 11 to go until election 2006. Yet despite these levels of support, the Liberals ended up falling out of favour over the course of their time in government and - most obviously and importantly - during the campaign itself. The sponsorship revelations had something to do with this, of course. But so did other events (the CPC performance in the campaign, Income Trust, fall-out from the budget vote/Belinda crossing, etc...) Harper, written off as an ineffective leader with no hope or future by the talking heads in February 2005, scored a remarkable victory in January 2006.

That 2005 Ekos poll, as a predictor of future electoral outcomes, proved a complete red herring. Yet who could be surprised at this. The poll itself only puported to ask who voters would support if the election were held today. BUT THE ELECTION WASN'T HELD THAT DAY. Nor is our next federal election expected for some time.

Why this patently obvious point is consistently overlooked by every pundit who puts these polls forward as news escapes me. It is analogous to a sports commentator drawing conclusions as to the final outcome of an NBA playoff game at the end of the first quarter, or on the likelihood of an NHL team hoisting the Cup at the All-Star break. While the score and play up to this point is clearly relevant, it alone does not dictate the future.

But no matter. The same folks forecasting the inevitability of a Harper majority next Spring will no doubt act as if they knew all along that Harper was too "extreme" for Canada if one of the Liberal leadership contenders emerges above the fray and sparks them back to office. Or not. And we casual bystanders will merely look on in feigned amusement at it all.

6 Comments:

Blogger Jason Cherniak said...

I'll admit to having given polls too much credence over my time following politics. However, I can at least argue that I am new to it all and had never seen a change in government before. I find it completely incomprehensible that people could be making exactly the same mistakes within a year of the roller coaster from June 05 - Jan 06.

7:30 PM  
Blogger Michael Fox said...

Polls make for fun entertainment, especially for political geeks. Aside from that, people get far too worked up about them.

7:54 PM  
Blogger The Tiger said...

If you wanted to make your point even stronger, James, you could have pulled one from mid-December 2005, when the Liberals were up at 40.

I trust that the leaders of the various parties in Ottawa know enough to take the long perspective.

7:54 PM  
Anonymous Anonymous said...

"Polls make for fun entertainment, especially for political geeks. Aside from that, people get far too worked up about them. "

Which is why Toronto Turd has posted his squeals of delight on several blog sites.

8:12 PM  
Blogger J. A. MacDuff said...

Jason - Incomprehensible, but every corner you turn around there is a pundit who argued adamantly that the Tories needed a new leader last summer, and now don't see how Harper can fail to win his majority. Wells is right: Amnesia as the journalist's best friend, I guess.

Ben - I didn't use a poll from mid-December simply because we aren't really on the cusp of an election call. But of course the fact that everything can change suddenly only bolsters the point.

12:35 AM  
Blogger JG said...

Well, at least part of the problem is that journalists are statistically illiterate, and they seem to think that inferring long-term trends on the basis of a handful of datapoints is valid.

6:49 PM  

Post a Comment

<< Home