Three parties entered the Quebec election with a chance to win, and while the PQ appears to be stumbling to victory, it’s nearly impossible to predict what will happen on Tuesday.
Of course, that’s not stopping anyone.
La Presse (thanks to CROP) projects a PQ minority, while the National Post (thanks to Forum) and the Globe (thanks to 308.com) project a majority. TooCloseToCall has the PQ with a 1-seat majority.
Even in the best of times, seat projections are an inexact science, but this is an election where they might not fare much better than the TSN playoff monkey.
Unlike Herman Cain, I’m not someone who distrusts the polls – but as the Alberta election showed us, shaky methodology and a volatile electorate can lead to journalists frantically re-writing their copy on election night. A robo-dialled Forum poll last week showed the Liberals riding a post-debate momentum wave to a 6-point lead over the PQ and an 11-point lead over the CAQ. That poll sticks out like an anglo at a PQ rally, but it’s still being included in the cumulative poll projections, and whose to say we won’t get another outlier in the final week?
Complicating things is the turnout issue. We know the PQ and Quebec Solidaire vote skews younger and young people are too busy smoking pot and listening to loud rock and roll music to bother voting…but are the polls factoring this in? Do the student protests change this? It really comes down to which party’s supporters are most motivated after an uninspiring August election, and I don’t think there’s a clear answer to that question. 308.com gives the Liberals an “advantage” for being in government, but this is based on a study of just 16 elections, and I’m not convinced it will translate through to a party that’s on the way out of power. And if it does, the other seat projections are under-estimating Liberal support.
Even if we had 100% confidence in the polls, transferring that confidence to seat projections is a recipe for disaster. Seat projections work reasonably well when there’s a good baseline to work off and relatively uniform shifts. Last year’s Manitoba election is a good example of this – with support levels only moving slightly from 2007, seat projections were spot on the money.
The problem with this election is that we’re dealing with a jumble of 3-way races and the rise of a new party, so there’s no baseline to work off. We can’t simply assume the CAQ will build off ADQ support uniformly, because this is a very different “third party” than the one Quebecers rejected four years ago, and they’re targeting a different type of voter.
So even when we’re treated to detailed regional numbers, there’s no way of knowing how that vote will transfer to individual ridings. Precise regional splits are a rare luxury, as many polls treat “Montreal”, “Quebec City”, and “Rest of Quebec” like three distinct and uniform regions. Anyone who has looked at riding-by-riding results in Montreal will know that’s certainly not the case.
My gut feeling is we’re heading for a Marois minority (and we all know how much she hates minorities!), but conditions are ripe for a surprise. Tune in Tuesday.
11 responses to “Your Guess is as Good as Mine”
I think you mean CAQ, not ADQ….
That did take me back to the previous election though…
Otherwise good analysis, and good quip at the end about minorities…
I run a new website, but the old one I ran, riding by riding, was tops in terms of riding by riding projections federally in 2011!
Despite that, yes, it is all guesswork, but it is what information you put in to your “guesses” that make them closer or further from the mark.
Regardless, I think political junkies go to vote projection websites, in part, for entertainment. If we, the projectionists, want to keep our readers, we need to be close to the mark, or our entertainment value fades, but at the end of the day I see it the same as buying a Lotto ticket – it gives you a chance to imagine a future that you find interests you.
And yet, Eric Grenier’s site continues to be picked up by the Globe and Mail, even though it has proven to be an unmitigated disaster time and time again.
I’d lean towards a Bloc majority. As you said, those vague regional breakdowns are misleading, particularly in Montreal. The PQ has made gains, and those gains are almost certainly not in Anglo ridings (though if you are using some sort of regional uniform swing, you’d predict PQ gains in Westmount).
And on the motivation question, PQ supporters are the most motivated (Forum has been tracking this). They may also get a last minute boost from the supporters of Quebec Solidaire or Option Nationale – Marois’ rhetoric of late has been to play up secession (in order to shore up her base).
Lastly, the tendency of Grenier’s model has been to underestimate seat results for rising parties. And it would make sense that this would happen since projection models have little idea of where gains are coming. That is probably good news for both the PQ (yes, the PQ is down in the polls compared to 2008, but their relative lead in the province has increased and that is what matters) and the CAQ.
Underestimated seats of gaining parties
Sask 2011
Newfoundland 2011
Ontario 2011
PEI 2011
Canada Federal 2011
New Brunswick 2010
Exceptions
Alberta 2012 (one of the few cases where his model’s conservatism inadvertently helped him)
Manitoba (where nothing really changed anyway)
So insofar as there is a consensus that this is a toss-up, these factors push me toward a PQ majority.
My model has a general history of over-estimating gaining parties after I began to add trendlines. I had the NDP at 113 in 2011.
There is a large bloc of undecided voters in Quebec. From looking at the polls, it is pretty clear to me these are voters torn between the CAQ and PLQ. If they swing heavily to one or the other, that party wins the election. If they decide “a pox on both your houses” and stay home, the PQ wins a majority.
How various vote projectors decide to distribute this bloc of voters will determine how right or wrong they are.
I’m not convinced the undecideds in Quebec are higher than in any other elections. There was a big Globe headline to that effect over the weekend, but the story said 10% are undecided and 20% “might change mind”, which is pretty standard for a final election poll.
There was one article this morning about how the undecideds traditionally break Liberal in Quebec but, again, it’s hard to say if that will still be the case.
308.com makes me pretty frustrated. I don’t want to throw stones, but seat projections based on provincewide polls, with so few data points, and with so many three-way or worse races are entirely useless and 308.com’s track record speaks to that.
Sloppy, sloppy, sloppy.
FiveThirtyEight it ain’t. I’m not sure why the Globe and Mail gives it their imprimatur.
Congrats on avoiding all the s-Quebec-ulation for 6 full weeks. For the first few days, all the ‘predictions’ got a bit on my nerves, however I got over it quickly: no one can predict most elections, and this one is harder than most.
This is an election where all three parties had a legitimate chance of winning right up until the debates.
Definitely fun to watch. And if the outcome is a PQ minority, the fun won’t end on election night.
[…] CommentsCalgaryGrit on Your Guess is as Good as MineCalgaryGrit on Your Guess is as Good as MineJason Holborn on Your Guess is as Good as MineCorey […]