Federal Election Polling Thread

[rubecube]

Originally Posted by EldrickOnIce

Yeah - I get what you are saying. And it probably makes sense of most levels. I just can't see it happening that way right after an election (given my limited understanding of the procedural process).
I see Harper trying his play first, then the GG possibly turning to the LPC after that. Obviously this delays actually governing, but I think it is the most likely way something like this occurring.
But just speculation, based on my limited understanding.

Yep, I think we're in agreement.

[rubecube]

I actually think a Liberal government, propped up by the NDP could be good for Canada in the long run because it would likely force the Conservatives to move to the left on their social policies. Quite frankly, I'm surprised they haven't done this already. Their base isn't going anywhere and the social issues are really the only thing stopping then from capturing all of the red Tories.

[Dogbert]

New Mainstreet poll: C - 37, L - 29, N - 24. Sample size of 5,197, or roughly 13 times that of the Nanos rolling polls.

They're getting another majority, aren't they?

[Looch City]

Originally Posted by Dogbert

New Mainstreet poll: C - 37, L - 29, N - 24. Sample size of 5,197, or roughly 13 times that of the Nanos rolling polls.

They're getting another majority, aren't they?

That'd be pretty sad if true...

[Drak]

Originally Posted by Iggy City

That'd be pretty sad if true...

Very. And no one should be in power for that long.

[PeteMoss]

Originally Posted by Dogbert

New Mainstreet poll: C - 37, L - 29, N - 24. Sample size of 5,197, or roughly 13 times that of the Nanos rolling polls.

They're getting another majority, aren't they?

Seems like an outlier poll... but who knows.

[EldrickOnIce]

Originally Posted by PeteMoss

Seems like an outlier poll... but who knows.

Sample size drops MOA to 1.36%, but this could be the 1 time in 20 it's well off.

[crazy_eoj]

Originally Posted by PeteMoss

Seems like an outlier poll... but who knows.

Nanos' numbers are generally the outlier. But they poll such a small sample size it's not hard to see why. They dominate the news because they do a poll every night, but it's almost insignificant because of the small sizes.

And they continue to consistently underestimate conservative numbers, as they have in the past.

Virtually every other pollster, with much larger samples, show the Conservatives moving into majority territory.

[rubecube]

Originally Posted by Dogbert

New Mainstreet poll: C - 37, L - 29, N - 24. Sample size of 5,197, or roughly 13 times that of the Nanos rolling polls.

They're getting another majority, aren't they?

I don't trust Mainstreet. They consistently have the CPC polling much higher than everyone else and they weren't in the field between Oct. 1 - 5.

Also doesn't help that they're run by Post Media.

[Senator Clay Davis]

The Nanos poll is of all voters, the Mainstreet is only of decided voters. The Mainstreet poll is also 5 days old and the Nanos is obviously a daily poll. Ipsos Reid has one out where they are essentially tied

http://www.ipsos-na.com/news-polls/p...e.aspx?id=7013

CBC poll tracker also has a tie

http://www.cbc.ca/news2/interactives...015/index.html

One thing that is obvious is the NDP is headed to third place.

[Blaster86]

As someone pointed out on Reddit, the Mainstreet poll has a sampling issue.

Does anyone else think it's weird that the CPC lead by 5% over the NDP and 6% over the LPC with 18-35 year olds?

Check out the sample sizes too: 436 for the 18-34 group vs. 2031 for the 65+ group.

That's not what's historically seen. In the last Ipsos poll, the LPC (34%) had an 8-point edge over the NDP in this group (26%), which was itself 6 points up over the CPC (20%). Forum's September 29th poll showed the NDP (37%) well in front of the CPC and LPC (25% apiece). Angus Reid also has the NDP up (27%) over the LPC (23%) and CPC (18%). It is extremely unusual to see the CPC leading the 18-34 demographic.
I think that Mainstreet has a sampling problem here, at least with this demographic. Young Canadians are notoriously difficult to get on the phone anyway, but in the unweighted sample this age group composed only 8.4% of Mainstreet's responses. Forum had a more respectable 17.8%; this group really composes 29% of adults (CANSIM 051-0001).

This reeks of a poll getting the numbers they wanted to see.

[polak]

Yeah, that's a good point. How are these polling companies supposed to get a hold of the younger voters? I don't know a single person my age with a landline?

Those also tend to be the most likely to lean left.

[nik-]

And be the least likely to vote

[EldrickOnIce]

It's been pointed out numerous times that almost all polls generally under predict CPC support. This may over predict, but...

[PeteMoss]

Originally Posted by polak

Yeah, that's a good point. How are these polling companies supposed to get a hold of the younger voters? I don't know a single person my age with a landline?

Those also tend to be the most likely to lean left.

It says they contact mobile and landlines in the poll summary. Now obviously the split between those matters.

[edslunch]

Sadly the over 65 group is much more likely to vote so these numbers may not be too far off. Where's that suicide booth gif?

[edslunch]

Originally Posted by crazy_eoj

Nanos' numbers are generally the outlier. But they poll such a small sample size it's not hard to see why. They dominate the news because they do a poll every night, but it's almost insignificant because of the small sizes.



And they continue to consistently underestimate conservative numbers, as they have in the past.



Virtually every other pollster, with much larger samples, show the Conservatives moving into majority territory.

One thing I noticed is that according to PollTracker Nanos uses live phone interviews while rest seem to all use interactive voice response or web surveys. There is a phenomenon of the "Shy Tory" where people are less likely to admit voting for conservatives. I wonder if the method of polling skews the answers?

[crazy_eoj]

Originally Posted by edslunch

One thing I noticed is that according to PollTracker Nanos uses live phone interviews while rest seem to all use interactive voice response or web surveys. There is a phenomenon of the "Shy Tory" where people are less likely to admit voting for conservatives. I wonder if the method of polling skews the answers?

That and they use a rolling three day average so if you get one bad day of data it skews your polls for several days in a row.

[MarchHare]

Originally Posted by edslunch

One thing I noticed is that according to PollTracker Nanos uses live phone interviews while rest seem to all use interactive voice response or web surveys. There is a phenomenon of the "Shy Tory" where people are less likely to admit voting for conservatives. I wonder if the method of polling skews the answers?

Nanos was pretty accurate overall in the last few federal electiond, predicting support of all five major parties within the MOE of the final pre-election poll.

Final Nanos poll - Actual vote share

2011 Election
CPC 38.7 - 39.6
NDP 30.5 - 30.6
Lib 20.9 - 18.9
Bloc 5.0 - 6.1
Grn 3.7 - 3.9

2008 Election
CPC 37.1 - 37.6
Lib 26.7 - 26.2
NDP 20.3 - 18.2
Bloc 8.7 - 10.0
Grn 7.1 - 7.9

2006 Election
CPC 36.4 - 36.3
Lib 30.1 - 30.2
NDP 17.4 - 17.5
Bloc 10.6 - 10.5
Grn 5.6 - 4.5

2004 Election
Lib 34.2 - 36.7
CPC 30.4 - 29.6
NDP 19.5 - 15.7
Bloc 12.2 - 12.4
Grn 3.7 - 4.3

Looks like Nanos's numbers are about as close to the actual result as can be reasonably expected. Keep in mind that any given poll only represents a snapshot in time, though. A poll published today will not accurately reflect the outcome of an election in two weeks. Based on the last four elections, however, Nanos's final poll should be a pretty solid indicator of the actual popular vote.

[EldrickOnIce]

^ Thanks. Nice to see those historical numbers.
Works for me then - Nanos as the authority it is.