Wednesday, February 16, 2005

Two recent polls

SES and Ekos both recently released national party preference polls. Interestingly, the SUN news headline on the SES poll was: Liberals sinking in polls. The headline from the Toronto Star, publisher of the Ekos poll, was: Liberals within range of a majority. You would think the polls were in dramatic conflict.

The SES numbers are Liberal - 38, Conservatives - 29 and NDP -17, with the Bloc at 45% in Quebec. Ekos has the Liberals with 40.2%, Conservatives at 26.5% and NDP - 18.7% while the Bloc is at 45%, overall very similar to SES.

The SUN report made much of a 10 point Liberal drop in Ontario from 54 to 44 from a previous poll in the fall, attributing it, without any evidence, to the controversies over the sponsorship inquiry and same sex marriage. It is almost certain, however, that the 54% number in the earlier poll was simply wrong. In the 2004 election, the Liberals actually received 44.7% of the vote, close to the figure the recent SES poll . My view is that polls that deviate very strongly from election results should be viewed sceptically. Rather than a dramatic shift, it is more likely that opinion in Ontario hasn’t changed much since the last election. News reports like the one in the SUN newspapers rarely contextualize poll numbers by comparing them to the last election.

It turns out that the Star’s headline that the Ekos poll implied a Liberal majority was probably right. However, this is also true of the SES poll, although just barely. Using the Ekos numbers and my forecasting model, I have the Liberals at 163 seats (Conservatives - 66, NDP - 25, BQ - 54), while in the SES they are at 156 seats (Conservatives - 82, NDP - 20, BQ - 50). A bare majority in a 308 seat House is 155.

In both polls what pushes up the Liberals are large apparent gains in Saskatchewan and B.C. - that is what puts them over the top. However, the polls suggest the Liberals would perform better (much better in the case of B.C.) than they have in elections in these two provinces over the past few decades. This seems on the face of it to be unlikely. I conclude therefore that despite these results an early election still seems likely to produce another minority.

No comments: