2011 Canadian Provincial Elections - Wrap-up phase.

[Hatman 🍁]

EPP normally does quite well, and I have a better track record in predicting Ontario than them.

[Foucaulf]

There's no point debating about qualitative v. quantitative when the quantitative model in question is very raw. At least Threehundredeight tries to apply meaning onto its arbitrary modifiers; Teddy reduces everything to a proportional increase and calls it a day. A serious mathematical model would involve regression analyses, error calculation and an actual distribution. If you're going to make one-off predictions, the difference between a holistic assessment and a weak model is negligible.

After witnessing previous elections (New Brunswick, Federal), I'm not sure if any projector's model can function properly in an election with destabilizing swings. The problem is that future Canadian elections will be filled with them (Ontario, Quebec, BC, Alberta...)

[Teddy (IDS Legislator)]

Have you ever seen my website? I never "reduces everything to a proportional increase and calls it a day"

To be blunt, after I'm right and you guys are wrong, you guys will continue to not care and imagine I was somehow wrong.

I was right for the federal election, but I gained a grand total of 5 readers. If you want to be wrong about the election, go ahead and be wrong. I've offered you proof, and both of you changed the subject, so I will no longer be responding to either of you on this matter.

When the BC election comes up, I will link you guys to the LAST provincial election, where an independent blogger confirms that I was the closest. Until then feel free to imagine that I'm wrong about everything.

[Filuwaúrdjan]

EPP normally does quite well

And the reason why they fycked up earlier this year was because they were gutless wrt Quebec and (I suspect) wearing rose-tinted specs wrt the entire Toronto metropolitan area.

[Hatman 🍁]

EPP normally does quite well

And the reason why they fycked up earlier this year was because they were gutless wrt Quebec and (I suspect) wearing rose-tinted specs wrt the entire Toronto metropolitan area.

Indeed. I made similar errors.

Teddy, what were your Ontario predictions for the federal election? I'd like to see them. I know I did better than EPP, but if yours are so great, please show them.

[mileslunn]

EPP generally seems to always overestimate Liberal support and this worked well in both 2004 and 2006 where they did better than the polls predicted, but not so well in 2008 and 2011 where the Tories did better than the polls suggested.  Actually Democratic Space was one of the closest to the correct results.  If anything it seems there has been a last minute bounce for the incumbent party each election, although not sure if it will happen provincially.  Either way other than PEI uniform swings don't really work in Canada like they do in Britain.  Uniform swings can give you the approximate seats each party will win, but not actually predict which ones they will win as strong/weak candidates can always help or hurt a party in certain ridings.  Also local issues tend to play a bigger role in Canadian politics than British politics.  You cannot use a uniform swing either for congressional elections in the US while for presidential, generally there all polls for all 50 states so you can use those which is not very practical on a riding basis.

[DL]

Even in the UK, the swings are not all that uniform. Last year some ridings had 11 point Labour to Tory swings while ridings right next door had just a 2 point swing and of course all the models go out the window in Scotland where the SNP is a factor and where the Tories made no gains at all. At the national level in the UK there was a mammoth Labour to LibDem swing and yet Labour actually GAINED several seats from the LibDems that they had lost in 2005 - so go figure.

[Filuwaúrdjan]

Uniform national swing was never as effective at predicting the outcomes in individual seats as people seem to think it was once, but the then that was never actually the point. The point was that you could look at the swing in the first few seats to come in on election night (and this was when counting took longer; usually lasting much further into the next day than has become normal in recent decades) and then work out whether there would be a Labour or Conservative Prime Minister. Individual seats would always swing in weird ways (there was a marked swing to Labour in parts Lancashire in 1959, for example), but the theory was that all of those individual and regional movements would cancel each other out. It was developed as much for television as anything else, and in a country with a notably stubborn electorate. It obviously makes no sense in a country like Canada, not even for its original purpose. It doesn't even make sense to include parties other than Labour and the Tories in it when using it in Britain.

[Smid]

I've put together the results of the 2007 provincial election:



As always, bigger version in the gallery.

[Teddy (IDS Legislator)]

http://calgarygrit.blogspot.com/2011/07/today-in-slightly-more-interesting.html
Brad Wall
Bigger than Jesus.

[RogueBeaver]

Genco is running as a PC against Sorbara in Vaughan per the Star . 308 marks that as safe Liberal. A question to Ontarians: is it Sorbara's personal popularity that keeps his seat safe? Especially given Fantino's federal landslide in May.

Also, 308 projects a Dipper victory in Manitoba, 33-22-3 despite losing the popular vote 44-40.

[Hash]

308's prediction record, however, is crappy even by the low standards of election predictions.

[lilTommy]

Genco is running as a PC against Sorbara in Vaughan per the Star . 308 marks that as safe Liberal. A question to Ontarians: is it Sorbara's personal popularity that keeps his seat safe? Especially given Fantino's federal landslide in May.

Also, 308 projects a Dipper victory in Manitoba, 33-22-3 despite losing the popular vote 44-40.

The last poll i saw was a dead even tie 44-44 but with the NDP at 50% in Winnipeg... that means generally the NDP should be the government... the Liberals will be lucky to hold on to River Heights and redistribution has killed Inkster. Plus the NDP was seen (from what i can see?) as having managed the recent flooding quite well overall.

Something interesting to snack on, some new democrats are saying the NDP SHOULD lose, in order to stave off a sask result... not keen with that idea but i see their points

[Hatman 🍁]

308's prediction record, however, is crappy even by the low standards of election predictions.

That's why I'm running my own.  I'm upset that someone gave him the Manitoba transposition numbers (for free), and won't share. But, I'm going to try and make estimates for my next Manitoba analysis.

Genco is running as a PC against Sorbara in Vaughan per the Star . 308 marks that as safe Liberal. A question to Ontarians: is it Sorbara's personal popularity that keeps his seat safe? Especially given Fantino's federal landslide in May.

Also, 308 projects a Dipper victory in Manitoba, 33-22-3 despite losing the popular vote 44-40.

As I discussed in my analysis of Ontario, no seat is really safe for the Liberals considering the ridings that they did the best in provincially are all held by opposing parties in the House of Commons. http://canadianelectionatlas.blogspot.com/2011/07/ontario-election-2011-prediction-mid.html

(Vaughan went Tory, Rouge River NDP and Cooksville Tory.)

For the record, I still have the Liberals winning Vaughan, but that can change. I'm going to be releasing my seat by seat popular vote predictions in my next analysis which will actually look at local factors which is being ignored by 308.

[Hash]

Well, the 308 guy has become something of a snob who thinks that elections are predicted only by universal swings and those myths and ignores basically things which actually count.

[Hatman 🍁]

Well, the 308 guy has become something of a snob who thinks that elections are predicted only by universal swings and those myths and ignores basically things which actually count.

Yup. It was funny how people were making fun of him at the convention. I guess he has the last laugh though, as he writes for the Huffington Post and Globe and Mail from time to time,

[DL]

Well, the 308 guy has become something of a snob who thinks that elections are predicted only by universal swings and those myths and ignores basically things which actually count.

308 is trying to apply a BBC-style "swingometer" to Canadian elections and its like trying to jam a square peg into a round hole. Canada has too many varying regional and sub-regional trends and individual riding anomalies for that to work the way it does in the UK (or for that matter in projecting states in a a US presidential election)

[RogueBeaver]

To be fair, no pollster predicted the NDP surge before it happened. Back to my original question: how vulnerable is Sorbara?

[Hatman 🍁]

To be fair, no pollster predicted the NDP surge before it happened. Back to my original question: how vulnerable is Sorbara?

My gut opinion is that's he not that vulnerable.  I do not expect him to lose, but wouldn't be that surprised if he did. How's that?

[Hatman 🍁]

Actually, Vaughan might be a potential target:

Liberal results:

2006 fed: 60%
2007 prov: 62%
2008 fed: 49% (-11%!) whereas the Liberals were down just 6% province wide between 2006 and 2008.

If The Liberals are down 11% province wide let's say, and add in an additional 5% such as what happened in 2008, and you're looking at 47% for the Liberals. Still a victory, but it could be more than just 5%.

What do you guys think? I'm trying to apply localized riding results to predict what will happen. Vaughan is a bit tricky, as it is trending heavily to the federal Conservatives, but I don't know if that will translate provincially. I guess we could look at regional 905 level polls.

[RogueBeaver]

2011 fed: 29%. That's certainly an ominous sign for Sorbara.

[Teddy (IDS Legislator)]

Remember they traded a locally popular Liberal for a locally popular Conservative.

Provincially you have a locally popular Liberal running against some guy.

You could shave 5-10% off the riding due to demographics (IE jewish voters and etc) but that needs to come off the calculated result. IE if you think it's 55L-40P then it's actually 50L-45P etc

[Smid]

Remember they traded a locally popular Liberal for a locally popular Conservative.

I don't have any deep insight, but I'm pretty sure I read somewhere that the Tory candidate during the by-election (who subsequently held the seat in the GE and is the current MP) was a high-profile, strong candidate. It may be a case of when the riding jumps Tory, it may move in a big way, but until then, it may resist a swing.

[Hatman 🍁]

2011 fed: 29%. That's certainly an ominous sign for Sorbara.

Also, not as relevant. I *would* be surprised if Sorbrara got less than 30% letalone less than 40.

[Hatman 🍁]

Also, the 2008 result was 15% higher than the provincial average. That might be a statistic to look at as well. If the Liberals get 31% provincially (last poll), than that means Sobrara will get 46%.

I think no matter how you boil it down, must calculations put the Liberals in the mid 40s in Vaughan. Which is interestingly enough, much lower than the 52% 308 gave them.