The Not So Triumphant Return Of Poll Soup
Warning - The following post contains graphic information and poll numbers that may not be suitable for some Liberals. Reader discretion is advised.
With that out of the way, let's recap the four March polls:
Ekos (Feb 24 to Mar 8, n = 2488 auto dial)
Angus Reid (Mar 8 to Mar 9, n = 1021 online)
Leger (Mar 7 to Mar 10, n = 2153 online)
Ipsos (Mar 7 to Mar 9, n = 1002 phone)
RUNNING AVERAGE (change since December in brackets)
CPC: 37.7% (+1.6%)
Lib: 25.8% (-2.7%)
NDP: 16.4% (+0.8%)
BQ: 9.7% (-0.1%)
Green: 8.1% (-0.4%)
As a programming note, the average above is based on all polls (not just those listed here), taking into account:
-pollster accuracy (based on provincial and federal elections over the past 5 years)
-a 14 day half life (so a new poll is weighted twice as heavily as a 2-week old poll)
As the graph shows, we've definitely seen a widening of the Tory lead since January. To pinpoint it, I shrunk the poll halflife to 5 days and retroactively ran week-by-week averages:
As you can see, the gap really widens around early-to-mid February. Your guess is as good as mine as to the cause. The current theory I'm working on places most of the blame on Charlie Sheen.
So what does this all mean? Well, my seat projection formula gives the Tories an 85% chance at a majority, and projects the following seat ranges (at 95% confidence):
CPC: 148 to 178
LPC: 53 to 79
Bloc: 49 to 61
NDP: 17 to 31
Although the national vote numbers are largely in line with the last election result, the Tories get into majority territory because their support is up in Ontario (which means a dozen extra seats) and down in Quebec (which means a few less) and Alberta (which means zero less).
Now before anyone panics, keep in mind this is based on the polls we're seeing right now. Polls of a relatively unengaged electorate. The vote numbers will move during a campaign. The last three times the government changed in Canada, the party taking power trailed at the start of the campaign.