Newfoundland election, Nov 2015

[RogueBeaver]

Coletto suggests that he's bleeding a lot more than other PCs for precisely that reason.

[Hatman 🍁]

Looks like VOCM wasted a bunch of money polling 4 safe Liberal seats. Not that there are many non safe Liberal seats to choose from.

But, running 4 telephone surveys amounting to 2000 cases in such a small geographical area must've been very expensive. This is why most riding polls are done via IVR these days.

4 ridings that should actually have been polled:

Ferryland
Cape St. Francis
St. John's Centre
St. John's East-Quidi Vidi

[DL]

But, running 4 telephone surveys amounting to 2000 cases in such a small geographical area must've been very expensive. This is why most riding polls are done via IVR these days.

It would be almost impossible to do IVR polls in provincial ridings in Newfoundland. These are tiny ridings that in most cases only have at most 8,000 eligible voters and the number of live phone numbers is so small that an IVR poll with the typical 4% response rate would barely get over 100 completes!

[Hatman 🍁]

But, running 4 telephone surveys amounting to 2000 cases in such a small geographical area must've been very expensive. This is why most riding polls are done via IVR these days.

It would be almost impossible to do IVR polls in provincial ridings in Newfoundland. These are tiny ridings that in most cases only have at most 8,000 eligible voters and the number of live phone numbers is so small that an IVR poll with the typical 4% response rate would barely get over 100 completes!

We're doing internal IVR polls right now, and I can say this is flat out wrong, but I would agree that 500 would be impossible using IVR. Obviously phone surveys are going to be better quality, but they're very expensive. And you don't need 500 cases to know that these four ridings are going to go Liberal.

[RogueBeaver]

CRA: 67/22/10. Final Abacus is Friday, dunno if Forum's popping in.

[DL]

But, running 4 telephone surveys amounting to 2000 cases in such a small geographical area must've been very expensive. This is why most riding polls are done via IVR these days.

It would be almost impossible to do IVR polls in provincial ridings in Newfoundland. These are tiny ridings that in most cases only have at most 8,000 eligible voters and the number of live phone numbers is so small that an IVR poll with the typical 4% response rate would barely get over 100 completes!

We're doing internal IVR polls right now, and I can say this is flat out wrong, but I would agree that 500 would be impossible using IVR. Obviously phone surveys are going to be better quality, but they're very expensive. And you don't need 500 cases to know that these four ridings are going to go Liberal.

When the trend is overwhelming enough - you don't need much of a sample. A survey of 100 voters is accurate if candidate A has a 50 point lead over candidate B - since even an n of 100 has a margin of error of plus or minus 10 points. If a seat was closely contested, you could not get away with such a small sample...and as I mentioned the populations of NL provincial ridings are TINY. Even doing a CATI survey is challenging when in most cases you may only have a couple of thousand phone numbers in the riding and you could literally dial every single number 20 times and be stuck with just 200 responses...In this day and age about 90% of the population won't even pick up the phone unless they can recognize the phone number on their call display! and let's not even get into the issue of the total impossibility of getting cell phones at the individual riding level - meaning that any riding poll is actually just a poll of old fuddy-duddies who still have listed landline numbers and may have virtually no one under the age of 35 unless someone lives with their parents and happens to pick up the old rotary dial phone in the parlour.

The only saving grace is that there is some anecdotal evidence that Newfoundlanders are more willing to respond to surveys than other Canadians.

[Hatman 🍁]

Yes, riding polls have many issues. News at 11.

Where is Al when you need him?

[Boston Bread]

Forum:
52 (-13)
29 (+8)
19 (+6)

http://poll.forumresearch.com/post/2431/progressive-conservatives-ndp-increase-vote-share/

Great poll!

[RogueBeaver]

Wonder if Abacus will find on Friday and if Forum does a last-minute Sunday poll. Might be shades of NB.

[Hatman 🍁]

post-debate bump?

[Filuwaúrdjan]

Where is Al when you need him?

Well I wasn't bothering much with this thread 'cos this election was looking about as competitive as the average Soviet one (apt given that Newfies first Canadian-era Premier was an unironic admirer of the USSR despite being a 'Liberal'), but, yes, beware of constituency polls etc.

[136or142]

Forum:
52 (-13)
29 (+8)
19 (+6)

http://poll.forumresearch.com/post/2431/progressive-conservatives-ndp-increase-vote-share/

Great poll!

If anyone cares, you'll remember my (approximated) prediction was
Liberal 50%
Conservative 30%
NDP 20%

Unfortunately, since this is a Forum "Poll", I'm now forced to predict the Liberals will get 65% of the vote.  

[136or142]

Forum:
52 (-13)
29 (+8)
19 (+6)

http://poll.forumresearch.com/post/2431/progressive-conservatives-ndp-increase-vote-share/

Great poll!

If anyone cares, you'll remember my (approximated) prediction was
Liberal 50%
Conservative 30%
NDP 20%

Unfortunately, since this is a Forum "Poll", I'm now forced to predict the Liberals will get 65% of the vote.  

Responding to myself

Forum polls are wrong even when they're right. For instance, the final Forum poll on the Canadian election was virtually dead on, but the election polls they had prior to that were almost certainly considerably off.  So, I think it's fair to say that when Forum gets the actual results statistically correct, they pretty much just got lucky.

[andrew_c]

Abacus (final poll):
64 (-2)
22 (+3)
13 (-2)

http://abacusdata.ca/final-nl-poll-liberals-heading-for-a-big-win/

[RogueBeaver]

Seats likely to survive the Gritnami.

[Hatman 🍁]

Not a bad analysis actually, and good to see a bit of geographic detail.

I also think Grenier's recent methodological change (taking into account the past 3 elections) is an improvement on his model.

[RogueBeaver]

Whackadoodle Cleary ad makes me wonder if a few screws are loose up there.

[Hatman 🍁]

My seat forecast: http://canadianelectionatlas.blogspot.ca/2015/11/newfoundland-and-labrador-election.html

[Panhandle Progressive]

An hour until polls close, follow live election results here:

http://www.cbc.ca/includes/provincialelections/newfoundland-labrador/dashboard/

http://www.thetelegram.com/News/Election-2015-29386

[RogueBeaver]

Grits leading in all 8 seats on the board.

[DC Al Fine]

Early days but the Tories lead in... seven seats. That's surprisingly not that terrible.

[RogueBeaver]

33-6-1 now.

[MaxQue]

Thought the NDP candidate leading isn't dreadful former leader Lorraine Michaels, but Gerry Rogers, the only other NDP incumbent running this time, in St John's Centre.

[RogueBeaver]

CBC projects Liberal majority.

[MaxQue]

Seems PC is becoming the Northern and Western St. John's Suburbs Party.