Monday, October 19, 2015

What if Deviation from Poll Numbers in Atlantic Canada Holds throughout the Country?

The Liberals are way outperforming polls in Atlantic Canada. What if they outperform polls by the same amount across English Canada?

Out of 260 seats
LIB - 163
CON - 78
NDP - 18
GRN - 1

Shocking, right? And consider that the above is assuming that over/underperformance is the *minimum* of arithmetic and proportional change.

Have a good election night! It's likely to be easy if you're a Liberal.


Anonymous said...

Wow, CBC is projecting a liberal victory!

Anonymous said...

:) .... Majority government!! Stable government for 4 full years! He is going to bring back our beloved Canada, God willing!

Anonymous said...

Nanos was stunningly accurate.....They projected the liberals at 39.8. I believe they underestimated the CPC by 1 point but otherwise, reasonably accurate.

Following is the final one

LPC 39.8
CPC 32.1
NDP 19.1
Bloc 4.9
Green 3.2

So Angus Reid got it all wrong for the liberals and probably overestimated the CPC support

Election Watcher said...

Maybe next time I should just project on Nanos and Forum, adding 1.5% to the Tories for turnout. Would have worked great...

Angus Reid was off. I feel a little bad for EKOS, since if they had broken out their last day as Nanos did, they would have done much better.

Francis Jumla said...

I expressed my suspicion about some of the polls that appeared to consistently overestimate the CPC vote. Turns out I was right about a possible majority government

CT said...

The thing is, overall. the Liberals didn't outperform the polls, they hit their marks. Innovative 38-30, Leger 38-30, Nanos 39-31, Ipsos 38-31, Mainstreet 38-33, Forum 40-30. An Abacus poll that, from the #s that were shared, clearly pointed to a similar lead. The polling showed a clear consistent picture-a 6-8 point Liberal lead. ANd the regional polling with decent sample sizes--the Ontario #s, the Leger Quebec #s proved very accurate as well (as opposed to a certain firm's small sample regional #s-if this election proves anything about polling in Canada it is that EKOS' non-Ontario regional #s are essentially meaningless).

Even the "outliers" weren't all that far off. The only real stumble was Angus' likely voting screen, which you hitched your star to to the detriment of your projection. Please explain to me why likely voting screens are not worth the trouble anymore, esp in fed elections in Canada.

So the polls were clear. The real surprise was the remarkable efficiency of the Liberal vote. Maybe there's a magic threshold in Canadian politics (38 percent?, somewhere in the high 30s) where vote efficiency accelerates sharply....

Election Watcher said...

I'll do a detailed post on this later, but:

- Kudos to Francis!

- The CPC vote was actually indeed underestimated by polls, just not as much as usual (1 point; my adjustment was 1.5 points). But the Liberal vote was underestimated as well, and the NDP and Green votes were overestimated by more than I thought they'd be. Perhaps a last-minute strategic shift.

- Crucially, polls were off in Québec. That makes a huge difference: I projected 26 seats for the Liberals, but on actual numbers, I would have actually overshot and projected 47! Jean Lapierre said this morning that all week, Liberals' internal number suggested that the best-case scenario was 25 seats, and on Sunday they bumped that up to 30. They actually got 40. The unexpected seats were pretty much the entire margin of their majority.

- I would have done better ignoring riding polls in many cases. In fact, overall, I think adding riding polls was counterproductive. Sub-regional adjustments, on the other hand, were helpful.

Bryan Breguet said...

Yeah I'm annoyed everybody is like: my god the polls nailed it. Sure the last minute, small sample sizes polls got the national numbers right. But not in Quebec. Not unless you only used the last Forum. And missing Quebec the way they did is major for projections.

In my model, I change Quebec and I get 160 LPC...

Election Watcher said...

Indeed, Québec and the unexpected collapse of NDP strongholds (e.g. St. John's East, Toronto--Danforth) seemed to be big problems. The former was due to polls, while the latter could probably not be predicted by any sensible model.

I think proportional swing would have helped in BC and AB, but would have been counterproductive in ON.

Anonymous said...

@ElectionWatch, correct me if I'm wrong but from the results that I'm seeing, the Liberals got over 1.6 million over the conservatives majority in the last election. Conservatives lost around 200,000 off their 2011 totals but of all these new voters only half a million seems to have come from the NDP. That is absolutely stunning because the Liberals have identified a ton of new voters and they followed the script of the Obama campaign, i.e. you cant necessarily change the conservatives but instead go find new voters and I have a feeling there were a lot of first nations voters, students, new canadians(if 1.6 million became new canadians over the past decade) etc. I'm not convinced that its just the tone of the conservative party that caused them to lose. Its just that these are similar sentiments you hear from our neighbors down south about not being conservative enough. I have a feeling that they are just unable to grow beyond their base. Sure, there are some switchers but there is more to it than that. If the Liberal party actually meets the needs of these constituencies, they would surely be able to win at least one more term, I think. I know a lot more things can change between now and then but I think if they make changes like mandatory voting/online voting(a near certainty imo)/ranked ballots etc, it will be so much better.

Election Watcher said...

It looks more like 1.1 million. There were 2.84M new voters, and the Liberal vote count went up by 4.15M.

We can't really presume that all new voters went Liberal. Surely, some voted Conservative or NDP, so there may have been more switchers than you think. Still, I agree that the most plausible explanation for these numbers is that a large fraction of new voters went Liberal.

Of course there are many reasons other than tone why the Tories lost. I think that nowadays, with voter allegiance lower than what it used to be, when you're in power for a certain amount of time, you almost automatically make enough people angry that you get thrown out. And when you pick as many unnecessary fights as the Tories did, that "certain amount of time" shrinks by a few years.

I'm not so sure about online voting. It'd be very nice to have it, but there are substantial security issues. Will Elections Canada be able to resolve them? And I'm a little worried about mandatory voting: great if it engages people more or encourages informed but lazy people to vote, but if it just brings in a bunch of apathetic low-information voters, the level of political discourse may further decrease.

Anonymous said...

You do have a point about mandatory voting resulting in people who are not engaged and low information voters voters but I also see it as a problem with a tone deaf 30% electorate too whereby 30% decide to choose Harper no matter what. They filter out messages that don't align with their worldview. When you have very educated folks like a Joe Oliver unwilling to speak out when you have an autocrat at the top, I wonder about the high information voter because they have the information but they are still muzzled. So I ponder if there is any difference. Because all i see is a conservative voter base which is highly motivated and just because they show up at the polls they are being rewarded as they know the other side doesnt care enough and if they are motivated, they try to make sure they dont show up by pursuing voter suppression tactics.

I do like the idea of ranked ballots however. Its better than proportional representation. My main concern about proportional representation is the proliferation of racist and xenophobic parties like in Europe and once they get into parliament, with their megaphones, they can create quite a ruckus. First past the post as it stands now is a huge problem if the conservatives ever win a minority/majority again because Stephen Harper used the modus operandi of incremental change to get people used to the idea that they arent as bad as people were making them out to be and once they got their full majority, we got to see them in action.

Election Watcher said...

Well, I think many voters of all stripes refuse to hear messages that challenge their worldview. If we had a two-party system, I'm sure that tons of people would vote for the left party no matter what. And then of course there are staunch Conservatives that are aware of their party's problems, but weigh certain issues enough that the CPC is still the best choice despite the problems.

I'm just not convinced that the opinion of people that don't even care enough to vote should count for as much as the opinion of people that take the time to cast a ballot.

I'm not a big fan of proportional representation either. I think we need a majority from time to time to move things along, and requiring a party to be the first choice of 50% of people seems too high of a bar. On the other hand, we should probably change our current system, which allows a party that's most people's last (mainstream) choice to form a majority. There are tons of ways to make this happen (ranked ballot is one of them) - it'll be interesting to see what we end up with.