June Poll Soup
Ekos (June 25-29, n = 2262 demon dialed)
Lib 32.2%
CPC 31.0%
NDP 16.2%
Green 11.5%
BQ 9.0%
Nanos (June 17-21, n = 781 telephone)
Lib 36.3%
CPC 32.2%
NDP 16.8%
BQ 9.8%
Green 4.8%
Angus Reid (June 17-18, n = 1005 online)
CPC 32%
Lib 31%
NDP 18%
BQ 11%
Green 7%
Ipsos (released June 20, n = 1000 telephone)
Lib 35%
CPC 34%
NDP 13%
BQ 10%
Green 8%
Harris-Decima (May 28-June 8, n = 2000 telephone)
Lib 35%
CPC 31%
NDP 15%
BQ 9%
Green 8%
OVERALL (change since May in brackets)
Lib 33.9% (-0.6%)
CPC 32.0% (+0.2%)
NDP 15.8% (+0.8%)
BQ 9.8% (+0.8%)
Green 7.9% (-1.9%)
ALSO...
-This CROP poll is spun as the end of the Ignatieff honeymoon in Quebec, but an 8 point gain vis-a-vis the Bloc isn't a bad way to end the honeymoon in my humble opinion. (And Leger has the Grits and BQ tied)
-There's a new site for polling info - threehundredeight.com (wherever did they get that name idea, eh?). There's nothing super high level as of yet, but it seems to be doing a good job posting poll results, and they do have a funky graph on pollster biases.
-Speaking of which, there's some interesting back and forth between Paulitics and Ipsos Senior VP John Wright.
-The usual Canada Day poll is out. In addition to the standard ("we love our country", "we're different from the US", yada yada yada), there's majority support for ditching the Monarchy - something I'd be completely onside with.
Labels: Polls
16 Comments:
Ignatieff is brutal. He has worked hand in hand with Harper to make Canada a more reactionary conservative country. Their latest joint venture to give mandatory minimum sentences for non-violent crimes is disgusting. I know Duceppe is a skilled enough politician to take the boots to the Count once the campaign gets going and I think the NDP have to be looking forward to going up against the most right wing Liberal leader in a long time. I've been sympathetic to the idea of strategic voting in the past but you could flip a coin between Iggy and Harper for all I care.
By Anonymous, at 1:13 a.m.
Better yet - why don't the NDP just go away.
After 75 years, it's obviously the CCF/NDP are not what Canadians want.
By Anonymous, at 6:42 a.m.
The usual Canada Day poll is out. In addition to the standard ("we love our country", "we're different from the US", yada yada yada), there's majority support for ditching the Monarchy - something I'd be completely onside with.
Not me. Harper already is presenting himself as a king. Don't encourage him, I say.
By Greg, at 7:54 a.m.
I'm not so sure about 308's polling bias numbers. The problem is that they are being compared to the average of all pollsters. That assumes that the average of all pollsters is the best predictor of election outcomes - a generally false assumption.
Angus-Reid was the closest pollster in 2008, for instance, because they were an outlier rather than close to average (they predicted high numbers for the Tories - actually that was the first time I was happily surprised by election results).
"After 75 years, it's obviously the CCF/NDP are not what Canadians want."
You can be influential without forming a government, or even coming close. Without the NDP the Liberals would tack to the center and something like universal healthcare would not have emerged in the 60's (it probably would have happened eventually). Similarly, the Reform party did poorly in elections, but pushed the clarity act, deficit reduction and opposition to constitutional accords into mainstream discussion when the big parties failed to do so.
By french wedding cat, at 9:53 a.m.
H2H - agreed re: bias. It's an interesting comparison that certainly matches what most people who follow these polls can intuitively figure out themselves. But "bias" probably isn't the proper term to use.
By calgarygrit, at 1:33 p.m.
I frame it more as a "bias" in their polling method, which isn't necessarily deliberate. For example, a polling firm that has managed to reach more young voters effectively (though internet or cell phone) will give better results for the NDP and the Greens. One that polls more heavily in the GTA might have higher Liberal numbers in Ontario, etc. etc.
And, no, it isn't scientific but I think it is a pretty good indication of trends.
By Éric, at 1:49 p.m.
"Ignatieff is brutal. He has worked hand in hand with Harper to make Canada a more reactionary conservative country. Their latest joint venture to give mandatory minimum sentences for non-violent crimes is disgusting."
No shit. Someone should send Ignatieff a few seasons of The Wire. In only a couple of years Canadian Liberals have gone from reform to reaction on crime and justice.
With all the glowing press, Ignatieff has received and the horrible headlines for Harper it is pretty telling that the Liberals have already stalled. The party learned all the wrong lessons from the Dion experiment and deserve another beating.
By Anonymous, at 2:57 p.m.
"I frame it more as a "bias" in their polling method, which isn't necessarily deliberate. For example, a polling firm that has managed to reach more young voters effectively (though internet or cell phone) will give better results for the NDP and the Greens. One that polls more heavily in the GTA might have higher Liberal numbers in Ontario, etc. etc."
That will not happen in any respectable poll because they weight for things like age or region. I mean there may be differing assumptions about demographics, but it isn't as important as in the US (where many polls are weighted by party affiliation).
The implication of using cell phones isn't that "you will get more young people", it is that you will get a more accurate sample of young people. For that to induce bias you have to convince me that the sample of young people with cell phones is politically different from those without. You might have a case, but it is a different kettle of fish.
Angus-Reid, Canada's biggest internet pollster (we are not talking about globe and mail type polls), for instance, has tended to have the highest numbers for the Conservatives. That isn't because "Conservatives are more likely to be on the Internet". I would guess it is because:
1. Being a Conservative is the least "socially acceptable" political affiliation. There are closet Tories that are more likely to identify as such in an Internet poll than on the telephone (particularly if the person calling is female).
2. Conservatives may do well among people that are only casually interested in politics. Actually I think this is plausible - Reform, the Alliance, the Ontario PC's and the Conservative party almost always outperform pre-election polling. You may pick those people up with Internet polls, which are less onerous, than with telephone polls, which are kind of a hassle (I was polled once).
My earlier point is that "bias" is relative to reality, not relative to other abstractions of reality. If reality has a conservative lean in Canada you will not get very far by assuming Angus Reid is an outlier (they were the closest in 2008 - and even they seriously under-estimated the Conservative result).
By french wedding cat, at 8:02 p.m.
Oh and God Save the Queen! With any poll on the monarchy you have to remember:
1. Treason is a crime
2. It is treasonous to oppose Her Majesty
3. Traitors should be hung
So, once we hang the annexationist rabble, you will see a resounding majority in favour of Queen and country!
And possibly a majority in favour of the restoration of Canada's proper flag, the red ensign.
By french wedding cat, at 8:08 p.m.
It isn't about whether one polling firm is an outlier while the others are normal, it is really just about comparing the polling firms to each other. For example, if Polling Firm X has the NDP very low but this is what this polling firm usually does, it isn't as much of a source of worry as when a firm that typically gives the NDP high numbers gives them a low one.
And as to polling firms taking into account demographics and the like correctly, something has to explain when a polling firm has a result that is very far from the reality. My calculations don't explain the disparity between polling firms, but it does present it.
By Éric, at 8:12 p.m.
"It isn't about whether one polling firm is an outlier while the others are normal, it is really just about comparing the polling firms to each other."
"
And as to polling firms taking into account demographics and the like correctly, something has to explain when a polling firm has a result that is very far from the reality."
So are you interested in comparing pollsters to each other or to reality - there is a difference, and your statements are conflating the two.
In Canada it is very hard to tell the difference, however, because each pollster has puts out so few polls. Those US numbers (eg. surveyusa's scoresheet) come from reams of state-level polls as well as national ones.
If Angus-Reid does 6 polls a year, and is the most prolific pollster simple random noise could easily explain why their results often lean conservative (this month they don't). Hell a week's difference in timing could explain the differences. I just don't think it is a particularly worthwhile enterprise.
A better and more productive approach would be to look at the actual methodology of the firms and figure out through abstract reasoning how the choices of pollsters will effect a poll result.
-Registered vs. likely voters?
-What proportions do they give to different demographic groups?
-How big are their regional sub-samples?
-Telephone, Internet or auto-dial?
By french wedding cat, at 1:41 a.m.
Oh and that paulitics-Ipsos debate represents everything that is wrong with Canadian politics... We are so concerned with framing events as a victory for our partisan preferences that the horserace becomes the story.
Polls are perfect for this. They have enough of a margin of error to produce the illusion of change. This, thankfully, saves downsized writing staffs at the Globe and Mail from having to write anything remotely substantive.
By french wedding cat, at 1:46 a.m.
As we get to know about Iggy, we find lots to not like.
By Anonymous, at 11:56 a.m.
I think we're headed towards a major political crisis when the Queen dies and it will put much of our last vestiges of monarchy into debate. How many people really want to see King Charles on the back of every coin or his face in the room of every municipal council chamber?
We may not officially separate from the commonwealth, but expect to see at least a major watering down of the monarchy's symbols in Canada's civil society similar to what happened in the 1960's. People have a hard enough time tolerating the power of the Queen now and I can assure you that they will be much less tolerant when the King is a scandalous buffoon like Charles.
By Spudster, at 12:25 p.m.
Any idea how big the undecided vote is?
By daveberta, at 12:48 p.m.
Nanos had 22% undecided in his...haven't seen undecided listed anywhere else.
By calgarygrit, at 12:59 p.m.
Post a Comment
<< Home