The election campaign that started out as a boring political version of Groundhog Day – essentially a repeat of the last two election campaigns here in Canada, has taken a sudden, unexpected turn. A series of polls in recent days have shown the New Democratic Party (NDP) – Canada’s perennial fourth place party in the House of Commons (behind the Conservatives, Liberals and Bloc Quebecois in number of seats) surging ahead in the polls. One poll even puts them firmly in second place, and NDP bloggers are squealing with glee as thoughts of Official Opposition status dance in their heads.
There is cause to take all of this with a large grain of salt. Polling in Canada is problematic, increasingly so. I have posted about this previously, and to avoid repeating everything in this post, I would urge everyone to read my earlier post on the matter, including the newspaper articles linked to in that post. When you have some of Canada’s leading, most respected pollsters admitting that the polls they are producing are close to worthless, it’s rather difficult to take any of these election polls that seriously. To quote from my earlier post:

Andre Turcotte, a pollster and professor at Carleton University in Ottawa stated that the various voting intention polls “don’t really mean anything”, while veteran pollster Frank Graves of EKOS said “The way it’s working now is a real dog’s breakfast. It’s not working.” Allan Gregg, chairman of Harris-Decima admitted: “The dirty little secret of the polling business . . . is that our ability to yield results accurately from samples that reflect the total population has probably never been worse in the 30 to 35 years that the discipline has been active in Canada.”

The problems with polling outlined in that post and the articles it links to (again, please do take the time to read the articles) were reiterated just a couple of days ago in another article again by Joan Bryden (please note – this article, from the Winnipeg Free Press, is no longer available online. I have attempted to find another source for it, but with no luck):
Voters trying to make sense of the raft of contradictory daily public opinion polls during this federal election campaign are likely to wind up with a migraine.
So who among the plethora of pollsters should they believe?
“Nobody,” says veteran pollster Allan Gregg, an outspoken critic of his own industry and chairman of Harris-Decima, which conducts polls for The Canadian Press.
As far as Gregg is concerned, the election campaign has magnified problems with political polling: methodological issues that are skewing the results of both telephone and online surveys; commercial pressures that are prompting pollsters to over-hype their surveys; and an unholy alliance with journalists who routinely misconstrue data and ignore margins of error.
The article goes on to repeat the what the earlier articles stressed: it’s becoming increasingly difficult for pollsters to get representative samples of the population. Online polls in particular tend to over-represent younger people who are more favourable to the NDP (but who are also much less likely to actually vote):
Jaideep Mukerji, vice-president of public affairs for Angus Reid Public Opinion — which has put NDP support at its highest level thus far, tied with the Liberals at 25 per cent — disagrees that online polls are less reliable. He points out that Angus Reid came closest to predicting the outcome of the 2008 election.
Nevertheless, Mukerji does agree that the idea of an NDP surge should be viewed with some caution. His firm’s most recent poll found that NDP support was the softest among the main parties. It could evaporate by election day, much as it did in 2008 when Angus Reid reported a brief surge in NDP support after the televised leaders’ debates.
Moreover, Mukerji points out that young voters may be more likely to support the NDP but they’re also the least likely to actually turn out to cast ballots.
The regional poll numbers are even more problematic due to their much smaller sample sizes. The margin of error in those samples is very high, and usually doesn’t get reported in the media. In some instances, because of the margin of error, any one of the three main parties could have been considered “in the lead” – the media simply chooses to focus on one (probably whichever will result in the most sensational headlines), ignores the margin of error caveat, and reports less than accurate information.
This morning, via Twitter, I found a very interesting blog post. Apart from the misleading poll numbers, the author also looks at how, in past elections, certain right-wing forces manipulated polls and campaigns to undermined the opposition parties:
The day you hear a Conservative suggesting that people vote NDP, “their ideological opposites”, is when you know there’s something else going on. They are trying to throw us off course. And it would appear that Jack Layton has taken the bait.
He will now pound away at Michael Ignatieff, in the same way that he pounded at the carbon tax, giving Harper a stronger minority.
This might read as being somewhat too “conspiracy theory” for some, but there is historical precedent here, as the blogger outlines.
Unfortunately, polling during election campaigns will not go away. I simply urge people to look beyond the headlines, find the margins of error, and any information about how committed poll respondents actually are to their choices, and then try to put everything in context. There are serious issues with political polling in Canada – our major pollsters admit this. The NDP surge may be real – or it may be the result of selective analysis and reporting of poll data by our media and its commercial interests. No one wants to report that another poll came out but who knows if there’s anything accurate about it when they can report that the NDP is on the verge of moving from fourth to second overall (or maybe even winning!). It’s fairly clear which slant will sell the most papers.