Captain's Quarters Blog


« Steal Big, Risk Little | Main | Air America: Goodfriend Says Piquant Strongarmed Him »

September 19, 2005
Liberals Back To 2000? Not Quite

A new Leger survey shows that the Liberal Party has fully recovered the standing it held in 2000, when it sailed to a majority government, despite the damage done this spring by the Adscam scandal. Martin has gained strength in Western Canada, a troubling development for the opposition Tories, whose numbers dropped by eleven points since their peak in the spring:

The federal Liberals had the support of 40 per cent of respondents in a new poll - virtually the same level of backing they received in rolling to their majority government in 2000.

The Leger Marketing survey, conducted Sept. 6-11, pegged Conservative support at 24 per cent, while the NDP stood at 15 per cent and the Bloc Quebecois at 13 per cent. ...

Marois said the poll revealed strong growth for the Liberals in Western Canada, including a jump of 16 percentage points in Alberta in two months and an increase of 14 percentage points in British Columbia.

In Ontario, the Liberals outstripped the Conservatives by a 46-27 margin, while the Bloc Quebecois continued its dominance in Quebec, leading the Grits by a 55-34 score.

The only odd part of the Leger survey comes from a "redistribution" of the 20% of their respondents that selected Undecided. Nowhere does the CTV article explain that process, but a look at the raw data makes it look rather suspicious. The Liberals gained eight points in the transformation (32% to 40%), while the Tories only gained four (20% to 24%). In April, the redistribution favored the Conservatives by a single percentage point, but here it favors the Grits by four.

How does Leger "redistribute" the undecided? It simply removes the numbers for those respondents who claim not to know who they support, who claim they will not vote, and those who refused to respond. It then "redistributes" the remaining responses and comes up with a number. In effect, Leger redistributed decided voters. Running the numbers through an Excel spreadsheet, the figures come almost but not exactly even with their claims. The Liberals get 40% of those who claim a preference -- a much more accurate way of describing the data -- but the Tories get 25%, not 24%, while BQ winds up at 12.5%, which gets rounded to 13%.

The raw numbers make more sense than the redistribution. The Tories lost seven points, which show up directly as seven points gained by the Liberals. BQ and NPD actually both lost ground. For Alberta, where the Leger survey claims large gains for the Liberals, they don't even publish the undistributed raw numbers as they did in the spring. They also only have a sample of 108 voters, hardly a reliable basis for making this kind of prediction. In four years of surveys, Leger has never used such a small sample for Alberta. The same holds true of their data for British Columbia.

Undoubtedly, Martin has recovered some of his support nationwide. However, the polling samples and the methodology of redistributing surveyed voters should cause Canadians to take this survey with a very large grain of salt.

Sphere It Digg! View blog reactions
Posted by Ed Morrissey at September 19, 2005 6:08 PM

Trackback Pings

TrackBack URL for this entry is



Design & Skinning by:
m2 web studios





blog advertising



button1.jpg

Proud Ex-Pat Member of the Bear Flag League!