Don't trust the polls: The Liberals are down in Quebec

[e-mail this page to a friend]

by L. IAN MacDONALD
National Post, Friday, September 5, 2008

"The Liberals' morale is at zero" announced a headline across the top of two pages in La Presse yesterday. And that was just the turn off the front, where the banner was a composite quote from Liberal insiders: "In Quebec, it's a disaster."

Stephane Dion, glancing through his press clippings, must have felt tempted to cancel his free subscription. If, on election night, he confounds predictions, he'll be entitled to a Harry Truman moment, famously waving an erroneous headline back at his tormentors.

But how can such a negative story run just two days after the release of a Strategic Counsel poll showing the Liberals in a very competitive second place to the Bloc Quebecois, at 26% to their 34%, with the Conservatives trailing at 23%? What gives?

The polite answer is that the poll's margin of error on its Quebec sample -- 6.3% -- essentially renders the numbers meaningless. The less polite answer is that no one in the Quebec political class believes it.

Never mind that Strategic Counsel is closely aligned with an Ipsos poll for Canwest-Global last weekend, which had the Bloc at 34%, the Liberals at 27% and the Conservatives a distant third at 21%.

In Quebec, the only authoritative polls are local, notably CROP and Leger Marketing. And CROP, in La Presse last week, told a much different story, with the Conservatives at 31%, the Bloc at 30% and the Liberals trailing at 20%.

It doesn't matter how accurate the pollsters are in the Rest of Canada (ROC). When it comes to Quebec, they should be taken with a very big buyer-beware notice.

It begins with the sample size. CROP, for example, polls 1,000 respondents in Quebec. Strategic Counsel polls 1,000 for the entire country, including just under 250 for Quebec.

But it isn't just the size of the sample, it's where the sample comes from.

Pollsters from ROC are susceptible to over-sampling the island of Montreal, a bastion of Liberal support. CROP, on the other hand, provides regional breakouts for the Rest of Quebec, a crucial battleground of 50 ridings, including the 418 area that includes Quebec City and the eastern part of the province.

The Conservatives established their 2006 Quebec beachhead in the 418, and now hold 10 seats there. The other 418 seats are all shaping up as competitive races between Bloc incumbents and the Conservatives. The Liberals, by their own admission, are not even in the game in the 418. Only in the Montreal region, and particularly on the island, are the Liberals in a strong position.

As the Conservatives contemplate the possibility of graduating from minority government, their road map to a majority lies through the ROQ as well as the 905 suburban belt around Toronto and the 519 in southern Ontario.

When the numbers from the Strategic Counsel poll are applied nationally, it's a 37-29 Conservative lead over the Liberals. But if the CROP numbers are substituted in Quebec (a step that statisticians would admittedly frown upon, since mixing poll data is generally not done), the Conservatives would be seven points higher in the province, and the Liberals six points lower. That would yield a national Conservative lead of around 39-27, putting Stephen Harper very close to majority territory.

It remains only for Dion to confound expectations. He did once before in the Liberal leadership race. Can he do it again?

 
  © Copyright 2006-2012 L. Ian MacDonald. All Rights Reserved. Site managed by Jeremy Leonard