Canada 2011 Official Thread

[The Vorlon]

It will be funny if voters reward Conservatives with a parliamentary majority on May 3rd -the exact opposite of Liberal hopes when they brought down this government.  

Is there any likelihood of this happening?  

I think the most likely outcome is a result more or less status quo.

There are a couple reasons to believe a Conservative majority may be possible.

1) - There are some signs voter turnout might be a bit better this election as the attitude "What the %^$% another election?" gets some traction - this might help the Conservatives.

2) - NDP leader Jack Layton (even though I personally think he comes across as a sleazy used car salesman) has a LOT stronger personal ratings than the liberal leader (Who has horrid personal favorability, trust, etc...)  To the degree that this widens the Conservative/Liberal gap (or narrows the Liberal/NDP gap, depending on how you want to look at it),  the Conservatives do better.  ie if the popular vote is Cons 40%, Libs 25, NDP 20, the conservatives get more seats than if the Cons still get 40% but the Libs and NDP carve up their 45% say 30%, 15%

3) Both sides have an issue that is semi-working - The Coalition mantra of an aloof, arrogant Harper has gained a bit of traction, but so has the Conservative mantra of a power hungry coalition desperate to do anything it takes to reach for the brass ring...  We will see which narrative takes hold....

In 2008, as the following graphic illistrates, the Conservatives also underpolled in every single poll, as well as having the Conservative/Liberal gap understated, so the polls, while a guide, have ben less than perfect in Canada.

[The Vorlon]

What's the absolute best case scenario for the NDP? I take it that even with Layton cashing in on being the most likable federal politician for the past 3-4 years official opposition status is way out of their reach?

The talented Éric Grenier calculated 44 seats. I suspect this is an understatement. The NDP can hope for 50-60 seats if there was a real collapse of the Liberal Party - mind you, this would almost certainly result in a 200+ seat majority for the Tories, so it would be a rather Pyrrhic victory.  

http://threehundredeight.blogspot.com/2011/03/february-2011-best-and-worst-case.html

Realistically, I have trouble seeing them gain much more than they have. They might hit 40 seats if the Liberals do badly, but that assumes the Liberals would be worse off now than they were in 2008 under Dion running on the Green Shift. Conversely, their floor is probably around 18-20 seats.

The "NDP best case" is really why Layton forced the election.

Ignatieff is so tragically flawed that a liberal decimation with the NDP rising as the "true opposition party" (ie setting up a two federalist party scenario, with the NDP as one of the final two...) is possible...  Not likely, but possible..

It would represent an historic breakthrough for the NDP.. I could happen if Ignatieff  were to seriously frack up his campaign.....

[The Vorlon]

The NDP will gain seats if the Liberals are weak, and the Liberals are currently weak. When it comes clear that Harper will win a majority, people will be "free" to switch to the party they "truly" support. Some of the NDP's best elections have been during Majority governments, because there is no pressure to vote for Party A to keep Party B out.

A clear and well stated point upon which you are correct.

[The Vorlon]

I take it that even with Layton cashing in on being the most likable federal politician for the past 3-4 years official opposition status is way out of their reach?

Really hard to see it happen...

The BQ, baring some really unforeseen game changer, is unlikely to get under 50 in Quebec, and in the absence of an near total Liberal meltdown, it pretty hard to paint a scenario where the NDP gets over 50...

[The Vorlon]

Per the Montreal Gazette, the most recent Quebec polling is downright disastrous for the Liberals:

Leger Marketing:
Bloc: 39%
Conservatives: 22%
Liberals 18%
NDP: 16%

CROP:
Bloc: 38%
Conservatives: 23%
NDP: 20%
Liberals: 11%

Leger's poll was a subsample of a larger national poll.  CROP's poll was Quebec-only.  CROP's poll also shows the Tories holding their own in the Quebec City region despite opposing federal funding for a new hockey arena, trailing the Bloc by 3.

Using a simple "uniform shift" model in Quebec based on the last election:

The CROP Leger poll projects:

BQ = 51
Tories 10
Dippers 2
Grits 12

The Leger Poll CROP projects:

BQ = 55 (+5)
Tories 12 (+2)
Dippers 6 (+5)
Grits 2 (-14)

The Liberals at 11% seems... deeply unlikely......

[The Vorlon]

I regret breaking the hackfest, but here's Nanos' poll. It's the first serious reliable poll out there since EKOS/HD.

Purgatories 38.4
Libs 28.7
NDP 19.6
BQ 9.1
GRN 4.1

Nanos was best in 2006, but Angus-Reid performed best in 2008.

For the (tiny) bit it is worth...

Punching the Nanos numbers into a regional uniform shift model suggests..

Tories:    148
Grits:        76
Bloq:         50
Dippers:    34

Why are we having an election again?

Nanos was (scary) accurate in 2006, out by a fair bit in 2008,

[The Vorlon]

Why are you having an election again...is there a serious chance that tories wouldn't be the largest party (out right majority or plurality) in the commons?

How would it work where they wouldn't be asked to form the government...and how stable/viable would that situation be?

In theory....

Tradition in Canada (indeed the whole Commonwealth) is that The Governor General asks the leader of the party with the most seats to form the Government... the GG could, at least in theory, ask somebody else....

There is also precedent (not in Canada but other British Style democracies) for a so called vote of "Constructive Confidence" where multiple parties vote to create a government out of multiple parties...

In theory... The Grits, Dippers, and Bloq could do this.

In 1980 the Grits under Trudeau and the NDP under Broadbent formed a defacto coalition government though it was never "officially" a coalition.

[The Vorlon]

Nanos day 2

Purgatories 39.1
Liberal 32.7
NDP 15.9
BQ 8.7
Green 3.7

Hell, that looks worse and worse.

Likely just noise - Tory gap actually grew in Ontario.

Projected Seats         
Tories   Grits   Dippers   Bloq
6           23   3           0
10           18   2   45
64           27   15   0
40           11   5   0
20           11   5   0
1           1   1   0
141   91   31   45

[The Vorlon]

If they're doing daily polls, then the numbers will bounce around wildly.

It's actually a three day tracking poll. so daily changes are based upon 1/3rd of sample - ie Daily sampe in Ontario is about 90 or so, Quebec 75 or so,  Atlantic Canada about 30 or so...

There is also the issue of not being demographically/age balanced within each subsample.

400 a day is barely enough to balance the sample nationally each day, but no where near enough for each region.

For example, Ontario may have too may men one day while Quebec might have too few.. the next day, the balance may shift, which will, in addition to (huge) random error, also cause substantial systemic error regionally.

We will (of course) look at the regional samples cause we are all junkies, but they need to be taken with a megaton of salt...

[The Vorlon]

I think you're underestimating the size of Nanos' Ontario and Quebec daily samples.  If he's polling 400 people per day, Ontario's should be about 120, and Quebec's 100.  FWIW, Nanos' Prairies (which includes Alberta) should have a daily sample size of about 80, British Columbia about 60 and Atlantic Canada around 40.  

Not correct actually.

Nanos samples about 400 perple a day but uses relatively mild screening questions to sift the sample down to "committed voters" - "very" roughly his equivalent of a "likely voter"

Fot the last two days these screens have reduced the original sample of about 1200 people down to 939 and 940 people respectively.

In Ontario, the three day samples were 279 and 281 persons respectively (or about 93 or 94 people a day)

In Quebec the sampes were 233 and 224, of 75 people give or take.

Only something on the order of 77% or so are deemed "committed" - hence the difference between the reported sample sizes.

The daily numbers are pretty well meaningless.... The trend line over a number of days might sort of kinda be a bit useful.. maybe.....

Nanos also has issues with sample balancing within each day.

He reports a leadership Index which he reports out of each daily 400 person sample (ie each day is a new 1 day poll)

On Monday Harper beat Ignatiuk by 84.8 to 43.4......  On Wednesday Harper won by 105.3 to 46.2

The gap expanding by 50% in two days seems... unlikely.....

Nanos also had a... not great year in 2008... though in 2006 he nailed it.....

[The Vorlon]

I love this thing:

http://predictor.hillandknowlton.ca/predictor/

Is it accurate?

No. Swings rarely work and work even less in Canada. If I recall correctly, this is from the same thing which predicted Mario Dumont would lose his own seat in 2008. So, yeah.

Uniform shift models "sorta kinda" work - They will get you in the right general ball park, but they are no replacement for riding by riding common sense.

[The Vorlon]

New Ekos poss also has Tories flirting with numbers  where a majority might begin.

we shall see, but Nanos and Ekos more or less agree on trend lines, both showing a modest shift to the Conservatives..

Ekos, as a technical note is more or less Rasmussen in that they use AVR technology.  Unlike Rasmussen they DO call cellphones and their dual frame sample actually seems to work pretty good

It is (likely) just a happy moment in time, but Dual Frame IVR seems to generate pretty decent probability samples (at least in Canada) at least for now.

[The Vorlon]

The regional samples are all over the map.

Getting much closer in Ontario, but Conservatives surging just about everywhere else.

The top line numbers on Nanos and the trendlines basically agree with the Ekos polling.

Ekos is using a much larger sample of about 3000 or so and as a result the regional breakouts are less like to be statistical noise.

[The Vorlon]

Nanos: - April 13th, 2011

Tories 39.9% (+.2%)
Grits 30.4% (-.8%)
Dippers 16.3% (-.5%)
Bloq 9.1% (+1.3%)

A couple notes:

Nanos calls during the day as well, so this sample has "about" half of 1/3rd of the sample being post/during the debate, but as of now.. it's steady as she goes.

A good Tories sample drops for tomorrow, so expect bit of a Grit bump in the April 14th result.

The big Bloq jump, at least part of it is just that a very good for the Liberals/Bad for Bloq sample dropped off the rolling three day total in Quebec.

The French debate is tonight - if, as expected, The BQ leader mops the floor with the other three, we may see both a bump to the Bloq and, likely, a bit of a migration to one of the Federalist parties. - Expect Layton to be hammered by the Bloq re Bill 101 as that is where the Bloq sees soft federalist votes as beng up for grabs.

[The Vorlon]

Does the Bloc actually win any votes from Anglophones?

A fair number actually.

The Bloc is a smart tactical vote for a Quebec citizen. -

Quebec is subsidized by the rest of Canada by roughly 11 billion dollars per year, or about $5500 per family of 4 per year.

Quebec uses this transfer to maintain some of the most generous social programs in Canada (cheaper post secondary education, state funded child care, etc...)

If you view elections as a game where you want to "win" (ie derive the greatest benefits from the system relative to your tax inputs) - Then the Bloc is a smart vote.

[The Vorlon]

More Polling.

Tories destined for a majority: Pollster
http://cnews.canoe.ca/CNEWS/Canada/CanadaVotes/News/2011/04/12/17974001.html

Compas:

The COMPAS poll surveyed 2,300 voters by telephone April 6-11, and is considered accurate within 2.1 percentage points.

Conservative 45
Liberals 24
NDP 16
Bloc 8
Green 7

FWIW (not much) This translates seat wise to:

Tories 178
Bloc 52
Grits 52
Dippers 26

Full poll release:

http://www.compas.ca/data/110413-NationalElectionPoll_Prt1-VoteIntentions-EPCB.pdf

Leger Poll

Leger polled 1,037 people selected from its online panel of 350,000 Canadians. The pollster only quizzed those who said they watched the debate. The margin of error is comparable to 3.1 percentage points, 19 times out of 20.

The PM = 37%
Iggy = 21%

Who won the debate?: Angus Reid Says:

http://www.vancouversun.com/news/decision-canada/Four+debate+viewers+think+Harper+poll/4605023/story.html

Harper 42%
Layton 25%
Iggy 23%
Bloc 2%

Full Release:

http://www.globalnews.ca/decisioncanada/POLL+Debate+fails+change+voters+minds/4604903/story.html

Bottom Line.....

Layton did very well
Harper did himself a modest bit of good
Iggy did himself a modest bit of harm...

These are the findings of an Ipsos Reid poll conducted exclusively for Global Television prior to and immediately following the English-language leaders’ debate on April 12, 2011.

The pre-debate survey was conducted among 2,615 English-speaking Canadian adults. The post-debate survey was conducted among 2,365 English-speaking Canadian adults who watched the debate.

Respondents were selected via the Ipsos I-Say Online Panel, Ipsos Reid's national online panel.

The sample was drawn from a pre-recruited panel of over 6,589 voters. Weighting then was employed to balance demographics and ensure that the sample's composition reflects that of the population of English-speaking Canadian adults according to the latest Census data and to provide results intended to approximate the sample universe.

EKOS Poll

http://www.ekospolitics.com/wp-content/uploads/full_report_april_13_2011.pdf

 National federal vote intention:
¤ 33.8% CPC
¤ 28.8% LPC
¤ 19.1% NDP
¤ 9.0% Green
¤ 7.8% BQ
¤ 1.5% other

[The Vorlon]

I say this in response (that potentially could make me look very foolish) to looking at Vorlon's estimated seat totals a couple posts above showing the Grits and Dippers both with 52 seats.

Is it at all possible to see a situation where the NDP, akin to UK Labour, completely over takes the Liberals as the natural alternative to the Tories...rendering Libs sort of a rump in future parliaments?  Is that how people see this eventually playing out, or are things (as I'm guessing someone or some people will say) so fundamentally different that such a thing cannot, in all likelihood, happen.

That projection with the Bloc and Grits both at 52 seats was based SOLELY on the one COMPAS poll, and based upon a uniform swing model applied region by region. (it is NOT my projection in any way)

I think this poll is an outlier on the high side for the Tories, I think the Ekos poll is a outlier on the low side for the tories.

There is a scenario wher the Implosion of the Grits/Rise of the Dippers takes place.

IF the country comes to the conclusion that a Tory majority will happen... and....
IF the more socialist wing of the party loses all faith in Iggy.. and ...
IF the red wing of the Liberals really likes the NDP leader..
then there is mass Grit defection to the Dippers and the NDP is the new #2 behind the Tories....

The reality on the ground is very hard to achieve however.  The NDP's vote distribution is very inefficient.

For example, if the Bloq bets 30% in Quebec and the three federalist parties spit the remaining 70% equally.. the  Liberals get 17 seats, the Tories 13, and the Dippers just 4....

In Ontario if there was a three way tie in the popular vote, the Grits would get 40 seats, the Torries would get 42, and the Dippers just 24...

The math on the ground is really hard to put together for the NDP to beat the Grits in seats....

[The Vorlon]

Tories steady, Grits drop, dippers on a roll.....

[The Vorlon]

      Tories   Grits   Dippers   Bloq
14-Apr-11   Nanos   38.7   28.8   18.6   9
13-Apr-11   Forum   36   25   22   7
14-Apr-11   EKOS   35.3   27.8   18   9.6
13-Apr-11   COMPAS   45   24   16   8
13-Apr-11   Innovative Research   39   28   17   9
10-Apr-11   Harris Decima   40   28   15   8
   
               

http://www.nanosresearch.com/election2011/20110414-BallotE.pdf

http://www.hilltimes.com/dailyupdate/view/way_open_of_harper_majority_forum_research_nationwide_poll_04-15-2011

http://ipolitics.ca/2011/04/15/debates-hold-little-sway-on-voters-ekos-research/

http://www.compas.ca/data/110413-NationalElectionPoll_Prt1-VoteIntentions-EPCB.pdf

http://www2.macleans.ca/2011/04/12/harper-still-far-from-majority-poll/

http://www.harrisdecima.ca/news/releases/201104/1117-conservatives-regain-double-digit-lead

[The Vorlon]

Leger/QMI says:

Tories 38
Grits 26
Dippers 22

http://www.torontosun.com/news/decision2011/2011/04/17/18028066.html

The poll was conducted April 15 to 17, after last week's leaders debates. Leger surveyed 3,534 respondents selected randomly from its online panel of more than 350,000 Canadians. The pollster says results would be accurate to within 1.7 percentage points 19 times out of 20 for a similar-sized group selected randomly from among all Canadians.

[The Vorlon]

Probably an outlier like that COMPAS junk earlier, albeit it one which will make for some good headlines to give the stupid media some money.

I will say this one more time...

The regional breakouts on these polls are JUNK.

Lets look at NANOS...

The topline horserace number has been amazingly stable.. The Tories have a 10% lead give or take a bit of statistical noise..

Yet the regional samples have shown the Tories anywhere from up 18 to down 10 in Atlantic Canada..

In Ontario the Grits have led by 7 and trailed by 18...

The list of huge and irrational swings goes on and on..

What seems more likely to you... That we actually do have all these huge regional swings that all magically exactly cancel each other out at the National level, so that the national numbers remain stable, or that these regional "swings" are simple the outcome of pushing a too small (and unbalanced) regional sample to places it statisically has no place going?

Bad Journalism combined with stupid readers (if you ask me)

[The Vorlon]

I will say this one more time...

The regional breakouts on these polls are JUNK.

I agree with you, with a few exceptions.  Some of the Quebec-only polling by the likes of Leger is usually reliable, as are the Ontario subsamples if they are large and balanced enough.  All the rest, especially subsamples of Atlantic Canada and Alberta and/or the Prairies, often have very high MoEs, sometimes double-digits.

I agree that the Quebec only polling is ok, because the design their entire sample design to sample, well, Quebec....

The problems with the regional subsamples goes beyond mere margin of error...

For a poll to even have a chance to be right (other than by pure luck) it needs to be demographically balanced so the makeup by age, gender, education, income, race, etc matches the target population...

The nightly subsample in the Atlantic provinces is what... 30-35 people... you just simply cannot balance all the above factors in a 35 person sample.. it's just impossible...

In Ontartio where your subsamble is 300 ish over 3 nights you have a bit of a chance, but unless you specifically stratify within the subsample (which NANOS does not) your regionals are gibberish..

Again.. even taking MOE into account, can the Tories going from 12 up to 8 down to 8 up in a period of 2 weeks be explained?

There is more than MOE here at work.

CTV must love this poll.. it gives believable top line numbers and a regional story (real or imagined) ever night....

[The Vorlon]

the NDP is 2nd place in Quebec in every recent poll.

EKOS now has the NDP at 20%, and it hasn't been a "pro-NDP" pollster. Abacus also had us at 20%, Forum Research and Leger both have us over 20%, and while it's not on Wikipedia, according to threehundredeight, Environics has us at 20% too, but maybe they got it confused with EKOS.

Environics have the poll on their website: Con 39, Lib 24, NDP 22, BQ 9, Green 6

Separate figures are given for Ontario and Quebec, but not elsewhere. Ontario: Con 39, Lib 33, NDP 23, Greens 6. Quebec: BQ 37, NDP 26, Con 18, Lib 14.

Environics Poll

http://www.environics.ca/news-and-insights?news_id=61

thx

Abacus - Internet Poll

Tories by 8

http://abacusdata.ca/wp-content/uploads/2011/04/Ballot-April-18-2011.pdf