Federal Election Polls

[cybercoma]

I think in your assumption you are assuming that the voters as part of N know what's going on day to day. I don't think so.

40 out of 1200 people would be enough to make a sizeable difference. I'm sure 3% of respondents know what's going on.

[angrypenguin]

40 out of 1200 people would be enough to make a sizeable difference. I'm sure 3% of respondents know what's going on.

I eat my words. Thanks again for your analysis. So I gotta ask you, if you were to pick on one pollster org, what would you pick?

[cybercoma]

I eat my words. Thanks again for your analysis. So I gotta ask you, if you were to pick on one pollster org, what would you pick?

I would never pick one. It's always better to have more people in the field polling, so we can analyze them all and better understand what works and what doesn't. They've all got strengths and weaknesses.

[ScottM]

Nanos hasn't seemed erratic to me, maybe I didn't notice. Can you point to an example?

I think erratic was probably a poor choice of words. Noisy might be better. Their trend lines for the Liberals and NDP have been pretty steady, or distinctly show the increase and fall. The Conservative numbers have bounced around a bit, though. I guess I know that a nightly sample size of 400 will produce that, but I guess it just unnerves me. I'm too much of a numbers nerd to be able to handle that sometimes. Lol

Even so, I think we're in agreement on the larger point. Someone is wrong with the numbers they're putting it out, and we might not know who for another 13 days.

Edit: I'll give an example of the movement for the Tories. They were at 28.9 on September 25 and 33.0 on the 28th. That seemed like a lot of movement in only three days. Just a few days later, they were back down to 30.5. It's not necessarily inaccurate, but it can seem unstable in short bursts. I think I probably get too caught up in that sometimes.

Edited October 7, 2015 by ScottM

[cybercoma]

Scott, I'm glad you point out how the Conservative numbers are less consistent than the others. This actually supports the theory we've been discussing.

[ScottM]

Scott, I'm glad you point out how the Conservative numbers are less consistent than the others. This actually supports the theory we've been discussing.

I'll add this: When you take the polls as a whole, it seems to me that the Conservatives' numbers are where most of the disagreement among the all the pollsters, i.e., not just Nanos, is. If the shy Tory effect is playing a role in the Nanos numbers, and my theory that incumbency is worth a couple of points that I've floated around based on polling from 2004-2011 is accurate, Harper might surprise again like he did last time around.

https://en.wikipedia.org/wiki/Opinion_polling_in_the_Canadian_federal_election,_2004

https://en.wikipedia.org/wiki/Opinion_polling_in_the_Canadian_federal_election,_2006

https://en.wikipedia.org/wiki/Opinion_polling_in_the_Canadian_federal_election,_2008

https://en.wikipedia.org/wiki/Opinion_polling_in_the_Canadian_federal_election,_2011

It might be worth adding that while the Liberals were underestimated in 2004 and 2006, the Conservatives appear to have been slightly more so in 2008 and 2011. That may support your theory more. If it does, should we expect it to be more pronounced this time?

Edited October 7, 2015 by ScottM

[cybercoma]

It might be worth adding that while the Liberals were underestimated in 2004 and 2006, the Conservatives appear to have been slightly more so in 2008 and 2011. That may support your theory more. If it does, should we expect it to be more pronounced this time?

Not necessarily because I think the impersonal polls are controlling for that effect, hence the 10-11 point swing.

[cybercoma]

Oh and this is all just theorycrafting by the way. I'm really not sure why the numbers are so far out and probably won't know until Nanos himself does a postmortem.

[ScottM]

Oh and this is all just theorycrafting by the way. I'm really not sure why the numbers are so far out and probably won't know until Nanos himself does a postmortem.

I understand. I'm doing the same thing. It's easy to get so wrapped up in this stuff that we try to explain what happened before it might not even happen. It can be great fun though.

[cybercoma]

Nevertheless, I believe there are strong indicators that social desirability bias is at play here, so I'm standing by my seat prediction numbers in the other thread for now.

[CITIZEN_2015]

Liberal Lead Narrows to statistical tie again (Nanos, October 7th, 6:00 am)

Liberals 34.3%

conservatives 32.1%

NDP 23%

http://www.nanosresearch.com/library/polls/20151006%20Ballot%20TrackingE.pdf

National Ballot – The two way race between the Liberals and the Conservatives continues with the Liberals at 34.3%, the Conservatives at 32.1%, the NDP at 23.0% and the Greens at 4.6% nationally.

• Accessible Vote – Asked a series of independent questionsas to whether they would consider or not consider voting for each of the federal parties, 49.7% of Canadians would consider voting Liberal, 41.6% would consider voting Conservative, 40.3% would consider voting NDP, 23.5% would consider voting Green and 27.1% would consider voting BQ (Quebec only)

Edited October 7, 2015 by CITIZEN_2015

[angrypenguin]

I think at this point the ndp is toast

[ToadBrother]

Nevertheless, I believe there are strong indicators that social desirability bias is at play here, so I'm standing by my seat prediction numbers in the other thread for now.

I think you're giving a little too much weight to your theory, and a little too much weight to some polls that are actually fairly old. While I have problems with the small sample sizes of the Nanos nightly poll, I agree with Eric Grenier over at 308 that the Liberals are actually pretty darned competitive in Ontario, and that while the Tories will get the plurality, they won't get a majority, or even a non-confidence-proof minority.

[cybercoma]

I don't think I ever suggested that they would. I gave seat predictions in the other thread that show the Tories with a small minority. That was based on doing a riding-by-riding analysis with the best available information at the time.

New Abacus numbers out today: CPC 33%, LPC 32%, NDP 24%, GRN 5%

This is similar to the Nanos numbers as well, statistical tie between the LPC and CPC with the NDP well behind now.

[ToadBrother]

I don't think I ever suggested that they would. I gave seat predictions in the other thread that show the Tories with a small minority. That was based on doing a riding-by-riding analysis with the best available information at the time.

New Abacus numbers out today: CPC 33%, LPC 32%, NDP 24%, GRN 5%

This is similar to the Nanos numbers as well, statistical tie between the LPC and CPC with the NDP well behind now.

Which all suggests previous reports of an 8 point lead by the Tories was an outlier.

And honestly, this is what I actually expected several months ago; that it would be a tight race between the Tories and Liberals, and the NDP fading to more traditional level of support.

I have to say that the NDP supporters I've been talking to are decidedly less buoyant now than they were a couple of weeks ago. It very much looks like the Liberals have seized the ABH crown, and while that spells some trouble for the Tories in that it might reduce vote efficiency in key battleground ridings, in the end it's the Orange Crash that I don't the NDP expected at the beginning of this election.

[Smeelious]

The Signal this morning has the Libs and Cons tied at 132 seats...Oh wouldn't that be interesting.

[cybercoma]

Which all suggests previous reports of an 8 point lead by the Tories was an outlier.

It's not an outlier when several polls are reporting it.

[Icebound]

I eat my words. Thanks again for your analysis. So I gotta ask you, if you were to pick on one pollster org, what would you pick?

None.

Gallup in the USA has already admitted that polling is impossible in todays technological world... and they have stopped polling the Presidential race:

http://www.politico.com/story/2015/10/gallup-poll-2016-pollsters-214493

I don't know what percentage of people have call-display, but I am sure it is substantial, especially in cities. None of those people answer their phone to pollsters.

Polls are useless, and when a leading Polling Company admits it, then it probably is true.

[cybercoma]

Gallup is being ridiculous. Nate Silver's predictions have been incredibly accurate. Gallup just isn't a very good pollster.

[angrypenguin]

11% for the NDP as per Abacus? No way.

[ScottM]

11% for the NDP as per Abacus? No way.

They're at 24%. The 11% number is those who think they will win.

[Smeelious]

I'd feel really bad at this point if the NDP get reduced to a small handful of seats.

Especially since I feel like there should be more parties represented in parliament, not less.

[Vancouver King]

I'd feel really bad at this point if the NDP get reduced to a small handful of seats.

Especially since I feel like there should be more parties represented in parliament, not less.

Thank Mulcair"s principled but naïve stance on the niqab issue. That he failed to anticipate how his stance would play out in his own home province calls into question his judgement on other sensitive files.

[CITIZEN_2015]

Thank Mulcair"s principled but naïve stance on the niqab issue. That he failed to anticipate how his stance would play out in his own home province calls into question his judgement on other sensitive files.

Mulcair stood for principle rather than political manipulation of a sensitive issue which according to reliable reports has caused several attacks on defenseless women in both Montreal and Toronto who were wearing niqab or hijab. In my view it is a cowardly action by a complete coward to attack a defenseless women for what they choose to wear and equally cowardly to try to exploit a situation like this for political gains.

I believe 33% for conservatives versus 32% for the Liberals is the correct one. Two recent polls have suggested exact number but same polls showing Liberals rapidly gaining momentum. If this continues then by October 19 they may likely stand at 33% versus 35% likely enough to form a minority government.

Edited October 7, 2015 by CITIZEN_2015

[CITIZEN_2015]

EKOS just out and shows conservatives lead narrowed to under 2 points which is a statistical tie for first time. EKOS is the one who took many by surprise by showing a near conservative majority on September 29, showing the conservatives leading by 7 points.

AFTER 20-DAY LEAD, CONSERVATIVES NOW FIND THEMSELVES IN A STATISTICAL TIE A TALE OF TWO PROVINCES? [Ottawa – October 7, 2015] After holding a statistically significant lead for 20 consecutive days, the Conservative Party’s lead has shrunk to just under two points and the party is now statistically tied with the Liberal Party who, not even two months ago, were in danger of being squeezed out of what had looked like a Conservative-NDP race.

conservatives have overtaken both NDP and Liberals in Quebec but Liberals well ahead in Ontario now.

http://www.ekospolitics.com/wp-content/uploads/full_report_october_7_2015.pdf

And I am more confused than ever by these polls that one released this morning showing a narrowing Liberal lead (Nanos) and the other showing a narrowing conservative lead (Ekos) both used to show big leads not long ago for their statistically tied leading party and both released today!!!!!.

Edited October 7, 2015 by CITIZEN_2015