New Brunswick provincial election 2020

[Hatman 🍁]

the506 is crowdfunding a poll: https://www.gofundme.com/f/let039s-poll-new-brunswick

[Hatman 🍁]

Why are you all thinking Higgs has got this? Snap elections often backfire, spectacularly (though the public is often forgiving in minority situations). And we're in a pandemic to boot. From what I gather, people are pissed off at the prospect of having to go to the polls.

[Hatman 🍁]

yeah, we're spending way too much time talking about the NDP in this thread.

Never! Cait Grogan will bring them back to relevance!!!111

I must say, they've already nominated more candidates than in the last Newfoundland provincial election (and are polling just as well!), which was another example of the NDP not being prepared. And they still won 3 seats!

[Hatman 🍁]

Sad to see the NDP not run a single seat in the north. At one point, it looked like the party was going to make serious inroads there.

[Hatman 🍁]

Sad to see the NDP not run a single seat in the north. At one point, it looked like the party was going to make serious inroads there.

It's understandable and expected that the NBPA run no candidates in the fracophone seats, but any idea why has the NDP has decided to run no candidates in the north? It is clearly appears to be an informed decision rather than one decided by a lack of individuals expressing interest. Maybe it's tactical to protect the Liberal vote in their strongholds, just guessing here.

I doubt very much it was an "informed decision". Definitely lack of interest. If there was interest, no one past the vetting stage. The party has an interim leader, and was caught off guard, and is pretty much in shambles at the moment. Why *would* there be interest?

[Hatman 🍁]

As I announced on Twitter, we'll be going through on polling NB by tacking on a vote intention question on a recruitment survey. Since this is a crowdfunded poll, I'll make the raw data public, and you guys can parse through it to see if you see anything interesting going on at the local level (though, taken with a grain of salt considering the number of cases per riding will be small).

[Hatman 🍁]

I think the difference between our poll and the others comes down to how it was weighted. The sample itself leans pretty PC, but the regional weight really boosts the Liberals. The huge difference in how Anglos and Francos vote in the province means that being able to correctly weight for that becomes challenging. Getting it wrong can really throw off the topline numbers. I haven't seen that much swing toward the Liberals while we've been in field (though, there was a bit of an uptick today.. err Saturday), so that leads me to think the difference is in the methodology.

I originally had a language weight, but I tossed it, as it was weighing the Liberals even higher. I think having a regional weight is enough anyway. The language weight was based on the language the respondent took the survey in, but was weighted using census data to mother tongue/language used most often at home. It's not an apples to apples comparison, so it's problematic. Could be lots of Francophones doing the survey in English. Anyway, hopefully I'm not accused of herding, we're still an outlier (we'll see what Mainstreet's numbers are). It's still obvious from the regionals that the Tories have the seat advantage, and could still win a majority.

[Hatman 🍁]

The language weight was based on the language the respondent took the survey in, but was weighted using census data to mother tongue/language used most often at home. It's not an apples to apples comparison, so it's problematic. Could be lots of Francophones doing the survey in English. Anyway, hopefully I'm not accused of herding, we're still an outlier (we'll see what Mainstreet's numbers are). It's still obvious from the regionals that the Tories have the seat advantage, and could still win a majority.

I would think using the language in which they completed the survey would be a better indicator of who is actually Francophone.  Census data for Francophones in minority settings tends to overestimate numbers given the push by French community/cultural groups to get anyone who speaks the language to identify it as their L1 for funding (and other) reasons.  Not sure this would be the case in the Edmundston and North-West area because it is majority francophone, but in Moncton, might play a role.  People like to identify with their heritage group (Francophone here) even if if is not the language they most commonly use. 

I'm not sure, you may be right. But next time, I will try to add a question about mother tongue (same for Quebec) to ensure that a language weight will be doing the right thing. This poll didn't have the budget for any more questions, so I did what I could.

Speaking of which, I do suspect that Moncton will be closer than the poll suggests, as removing the language weight dropped the Liberal numbers there. Moncton, I think is hard to poll- looking back at our internal polling from 2018, our regional numbers were the most off there. And this time, the biggest shift in regional voting intentions (compared to 2018) is in Moncton. All the other regions have similar numbers to 2018. Again, this shift might just be explained on dropping the language weight, and not based on any actual shifts in vote intention.

[Hatman 🍁]

Hmm. Small sample size. Still making me nervous. But I suspect the difference comes down to how it was weighted.

[Hatman 🍁]

I've been panicking about our numbers ever since I saw them last night, and Mainstreet's numbers make me even more anxious. I've been trying to compare the two polls the best I can. I think the big difference is the under 35 numbers. Very hard to poll this group, so a crapshoot as to who is correct. We both have the Greens winning them, but we have the Liberals in 2nd and they have the PCs in second. This group has a poor response rate, so it gets weighted up 2 or 3 times, which really has an affect on your top line numbers.

[Hatman 🍁]

My prediction (this is more of a gut feeling than anything based on any modelling)
PC: 27
Lib: 18
Grn: 3
PA: 1

[Hatman 🍁]

When will we have most of the results? late tonight or later this week?

Tonight, possibly as late as midnight. I think ridings can't report results until all of the polling stations in that riding are closed, and so if there are long lineups, there may be delays.

[Hatman 🍁]

My prediction (this is more of a gut feeling than anything based on any modelling)
PC: 27
Lib: 18
Grn: 3
PA: 1

Damn, I should've advertised this more. Stick with what you're good at eh? Looks like I only got Miramichi wrong.  Sorry for botching the poll.

[Hatman 🍁]

Caitlin Grogan had the best NDP result with 6%. Who would've thought Quispamsis would be the best riding for the NDP I wasn't sure if her large online following would translate into votes. I guess it did a bit.

[Hatman 🍁]

NDP: Unequivocally a bad night, despite what some Redditors are saying. At least now they have four years to rebuild without worrying about a surprise election. My unsolicited advice: Talk about bread and butter issues more. Running on abortion and defunding the police, makes you seem like a meme, especially in a place like NB. When the right wing populists are talking about poverty more than you... it's not a good sign.

Unfortunately for the NDP, their membership has been reduced to a bunch of activists, and NB is not the kind of place where activist politics are going to be very popular. Most of their pragmatic voters are going Green now.  The activist lane they've gone down is anathema to the base they've have in previous elections (i.e. working class Saint John and Acadia). This time, they didn't even run any candidates in the north!

[Hatman 🍁]

My poll is giving me mental anxiety, so I have to keep analyzing it. Sorry, everyone.

But, I looked at the total error, and while it's bad on the PCs, we were actually closer to the other parties than Forum or Mainstreet:

Ekos: PC (-7); Lib (-3); Grn (+3); PA (+1); NDP (+2) - total error of 16
Forum PC (-2); Lib (-5); Grn (+6); PA (-1);  NDP (+2) - total error of 16
Mainstreet PC (-2); Lib (-5); Grn (+6); PA (-2);  NDP (+2) - total error of 17

What went wrong? It was in the age weighting. I ran our numbers with no age weight and got
PC 38, Lib 32, Green 15, PA 9, NDP 3. A total error of just 5! New Brunswick is rapidly aging, and this needs to be accounted for polling to be accurate.

[Hatman 🍁]

Looks like the Green vote share is well below what the polling indicated

You can almost set your watch by that for most elections.

All comes down to age weighting. Mainstreet weighted by Census proportions (<35 = 24%), so over polled the Greens more than we did (our <35 weight was 20%). In reality, with young people fleeing the province, their proportion of the electorate was probably much lower.

[Hatman 🍁]

Poll maps are up:
http://www.election-atlas.ca/nb/?e=2020

Saint-Quentin continues its streak of voting the opposite of Kedgwick

[Hatman 🍁]

Also a map by county and municipality would be interesting if someone could do one.

It will probably look like this (reverse the colours)