Election Eve Liveblog: On And On And On

A Rolling Final Day Chronicle

Thank Fuck This Is Over

So far this weekend we’ve seen last polls from Leger, Abacus, Angus Reid, MQO, Innovative Research Group, and Ipsos, plus tracking polls from the usual suspects. We’ve seen Frank Gravesi’ tweeted national number (46-37 to the Liberals) but no regionals yet, and we’ve seen a horrible tragedy in Vancouver that has me devastated and with nothing profound to say. (And honestly reminds me of how fundamentally inconsequential my ramblings are, and yet here I am indulging them some more.)

Over the course of today we’ll be getting the rest of the final polls, and so instead of one final big preview tonight, let’s just do it live. I’ll cover the battle for Ontario, what tea leaves say about Quebec, break down the West, and focus on the east, plus whatever else interests you guys - ask your questions in the comment, I’ll get to the good ones while I sit here.

Consider a paid subscription, if you can. This site, and this campaign especially, has been a labour of love, and I’m eternally grateful to everybody for reading this. That said, this is one of the most prolific sites for commentary that keeps everything in front of the paywall and a paid subscription is a great way to support me. All the content will always be free.

8:20PM - I’m done with this, I’ll tweet some reactions after all the polls are out but this day has driven me fucking batshit and I’d like to get properly hammered. Love my readers.

7:35 PM - The Liberals will project for a majority government in my model when I release it tonight. There is no mathematical possibility it will not after Research Co dropped their latest and nothing changed. Even if Pallas, Mainstreet, Nanos, and Liaison all saw changes that had the Conservatives at 70% of the vote nationally and in Ontario it would not be worth enough weight in the model to move it.

I’m just waiting for the final four (Forum doesn’t count) to drop before I issue the final projection, but right now the Liberals project for 183 seats and majority government, and they will project for majority government no matter what.

7:25 PM - there are some Spadina Strategies polls of the three Saskatoon ridings suggesting the Liberals are within 2 in two ridings. My response to that and all other glimmers of hope in Alberta and Sasky is to press X for doubt, but others are allowed to be less cynical than me.

7:00PM - EKOS is bad for the Liberals. Cost them two seats net in the model, as swings right compared to their previous poll in Ontario and BC lose them seats and they only gain one in Atlantic Canada. It’s still objectively a good poll - solid majority government if perfectly replicated - but it moved the model rightwards. Pallas, Research Co, and the trackers still to come.

6:10PM - We’re on standby for EKOS, apparently.

4:57PM - Currently just going through the NDP seats in my head again - I think I’m sticking with the core 8 (Boulerice, MacPherson, Green, Mathyssen, Masse, Blake, Julien, Kwan), but I’m worried about Gazan.

Less concerned about Davies mostly because of how pathetic and desperate his local campaign is.

4:12PM - The mood in Liberal circles, at least from what I’ve heard, is tense but optimistic. Everyone’s exhausted at this point, having worked for or helped Crombie and/or Carney’s leadership in many cases (and in a couple, both), but there’s no analytical or rational reason that’s been given to me that would concern them or me.

This isn’t to say they’re not worried - we’re Liberals, after all - but we can distinguish between legitimate concerns drawn from credible reasoning from Liberal fears because we all have undiagnosed anxiety.

3:55PM - I’m still here, I’m just waiting for pollsters to give me content.

3:20PM - The Conservatives have decided to launch a new video on their tough on crime agenda, including in the original caption a reference to last night’s tragedy in Vancouver, which is crass and contemptible but also bad politics. Getting the politics of tragedy right is hard, and the line of using it to make your point while not being seen to exploit it is hard. Here, the Conservatives didn’t need to say anything.

To whatever extent this might have moved votes, it would have been because this guy was a psychopath who has been in and out of prison, but it would have to be something where you’d let your media allies and friends make the argument for you. You can’t be seen to play politics with it, because you look like assholes. But even then, this doesn’t help them because it seems like it’s just some person fighting their demons who did this. This isn’t the Eaton Centre Boxing Day tragedy, which probably did help elect Harper. This is just tragic.

The problem for Poilievre is he’s seen as an unserious person leading an unserious party, and this is not going to help that impression. There’s nothing to say Poilievre ordered this, but it doesn’t matter - it’s his party, in his image. That anybody around him thought this was acceptable is a huge problem for anybody wanting to believe that the crisis we face would elevate Poilievre to meet the moment. He’s the same fucking weasel he’s always been, and his party should burn for it.

3:05PM - This is going to be one of the wonkiest answers I’ll ever give but it’s not new that IVR - or so called “robidiallers” - get an overly conservative sample of young voters. I remember this being true with ReachTel in Australia going back at least a decade, before any of the sort of recent talk of a less progressive youth was happening.

In general, I’d lean towards the demos of an online poll over the IVRs, especially when they’re directionally telling the same story. The Conservatives are uncharacteristically competitive with young voters, and competitive to win overall. The exact details aren’t super important.

2:41PM - I’m not sure what to make of this, and I’m certainly not modelling off of it, but I will say that there is a decent chance the talk of “tightening” - as ludicrous as it’s actually been statistically - will very much undo the tightening, as progressives who were going to vote NDP when they thought they were safe to come home.

There’s been a lot of ink spilled (in my view, rightly, even though I disagree with lots of it) on the chances of the NDP surviving through tactical voting, but I do think we’re possibly underestimating just how low the NDP could go in some seats - especially if there’s a recognizable hate figure of progressives there.

Also, I don’t want to drag another modeller and I think his work looks very reasonable in basically every other case but please stop quoting me Fournier showing Anita Anand in a tight race. She’s not. Something has gone wonky with that projection.

2:20PM - Expectations setting for Atlantic Canada will be important, so let’s deep dive into the four provinces, from least interesting to most.

PEI - If the Liberals are losing a seat on the island then there’s some big problems. I feel like if there was some great local crisis there we’d have heard about it by now, and I’ve heard nothing. Should remain a clean 4-0.

New Brunswick - if the Conservatives get 5 or more it’s notable. They have three fairly easily - Fundy Royal, Saint John—St. Croix, Tobique—Mactaquac - and Miramichi is a plausible fourth. After that, the pickings are fairly slim - maybe Fredericton? Cause Wayne’s not in trouble in Saint John and honestly if Dom is in trouble in Beausejour I think we might see flying pigs. If you’re the Liberals, 3 Conservatives is great, 4 is good, 5 is big polling miss territory, and 2 is 200 seats.

Nova Scotia - Here there’s more intrigue, just because there’s more competitive seats here. South Shore-St. Margaret’s doesn’t have an NDP candidate which helps us hopefully steal a seat from Rick Perkins, Cumberland Colchester is within a point in my model right now, but on a different night the CPC could be looking at gains in Cape Breton and even Central Nova. It’s likely the CPC are capped at 3, because this isn’t the kind of night where the Conservatives win or even come close, but whether Sean Fraser wins easily or not will be a bit of a tell. Right now I’ve got 2 Conservatives, and 3 is a mediocre but survivable result for the Liberals. Anything more and I’m worried.

Newfoundland And Labrador - this is the actual reason I feel the need to do this segment. I’m not even particularly scared, I’m just pretty certain this is going to go badly for us tomorrow. Now, “go badly” means 2 CPC seats, holding Central Newfoundland and gaining Long Range Mountains, but still. Central Newfoundland has a CPC incumbent and the loss of Scott Simms’ significant personal vote, while Long Range Mountains has a shit candidate and a campaign that is going shit, according to my sources.

In any other region I wouldn’t care - the math says both should go red - but this is Newfoundland. These are rural and regional seats where personal votes and candidates matter, and this is the kind of thing that sheer spreadsheet nerdery can never account for. Maybe I’m dooming too hard, but don’t be surprised by a 5/7 result for the Liberals - and don’t think it means anything about the national.

1:58PM - My main contribution to questions of an ethnic swing is to remind people that white people fucking suck at figuring out what’s noise versus what’s signal, and to remind them that the Liberals did really really really well in Brampton and Surrey last time - to the point where huge swings right and Liberal wins are both still possible, and arguably modal.

1:47PM - On the polls, and whether or not there’s going to be some form of differential response or Trumpian underpolling of conservatives or whatever - you can’t say never, but it’s remarkable how many people asking this haven’t been through the wars. We’ve been waiting for US polling misses to flow through to Canada for years now, and they haven’t, for a few reasons.

The first is structural - there’s just a greater proportion of the Canadian population in central Canada than in any combo of US jurisdictions. Getting Ontario right is nearly 40% of the battle, and from a competitive seats standpoint it’s even more. Throw in Quebec and you’re at 60% of the country before anything else happens, and our pollsters have a lot of practice being very good at polling those places, including this February.

Look, would I take the over on the CPC in SaskyToba and Alberta compared to their polling averages? Yes, and there’s a bunch of evidence of polling misses breaking their way out there. But that’s not really relevant to the national polling because, frankly, national numbers don’t matter. Oh no, we might not come close to winning a seat in Saskatoon, that’s a huge problem for my model that has us … not winning any seats in Saskatoon.

Secondly, it’s worth reiterating that we’ve done this discourse before in 2019 and 2021 and Doug Ford’s first win. Every big election is viewed as a test for the industry in this new, post-Trump 2016 world. It’s not the first time we’ve done this, and it’s been a waste of effort and energy every time.

The third thing is the sheer size of this lead. The Juno News “neighbour poll” - asking not who you’ll vote for but who you think your neighbours will vote for - only showed a 2% swing right from that. A 2% swing from the current numbers, evenly distributed, would still result in a Liberal majority. Double the size of the effect and it’s a strong Liberal minority where the NDP still have the balance of power and we get to do what we want for 2-3 years while there’s a leadership race and the next leader rebuilds the party (especially financially). And that’s even indulging the premise that it’s a better way of polling, which it’s not.

“Neighbour polls”, like every other form of pseudoscience nonsense, has a huge survivor’s bias to it. Advocates for it as a method only bring up the couple of times it was right, pretend it was the only method that got those times right (which definitely isn’t true about Brexit), and ignore the many times when the left either beats its polls or the polls are right. It’s a hack method trying to trick you with broken clock nonsense.

Anyways, there’s truly no reason to doubt the broad accuracy of the polls. They’re good at this! Very good!

1:26PM - One of the few things we do know at this point is that any idea of a Bloc surge is utter crap. Both Angus Reid and Abacus, the two pollsters who freaked out the weak kneed doubters earlier this week, have both shown robust swings to the LPC, as has Pollara in their daily tracking for Curse Of Politics. The idea of a post-debate surge mostly missed what I wrote about Saturday, which is that the Bloc are disadvantaged at this election by having to fight about things that Canadians care about as the issues Quebecers care about, and to which they have no answers.

The secretly far more interesting question about the Bloc isn’t how well they do - they’d need a serious polling miss to even scrape back to 25 seats, and the king of Quebec polling is more friendly to the LPC than the average right now so that seems unlikely - but how long Blanchet stays. If he continues to view the hallowed halls of Ottawa’s Parliament as a foreign one, I wouldn’t be surprised if he finds himself a PQ nomination for next year and finds his way into the provincial cabinet. Probably not right away, but I think there’s a non-zero chance that we’re in the final days of having to take Blanchet seriously.

The other point worth reiterating in Quebec is that they have an immense capacity to break late, as we saw in 2021. The evidence from the trackers isn’t there yet, and even then it’ll be shaky when we get today’s samples, but if there’s a polling miss in Quebec I’d probably bet on the Liberals beating their polls. Blanchet doesn’t trot out the artificial country line if he’s not seeing bad data.

Intro: Where Are We Now?

Yesterday I tweeted out that the Liberals were on 187 seats, and since then two things have happened - IRG released some Chinese-Canadian sampling that suggests those voters are swinging substantially to the right, and the Alberta polling got bad for the Liberals. That’s cost them two seats - Richmond Centre and Calgary Skyview - but not fundamentally done much to the projection.

Ontario’s been a bit weak in Abacus and MQO today, and so London West, Peterborough, and Thunder Bay Rainy River all look more lineball than they did. However, there’s every reason to think given what Frank has tweeted that his numbers just might see a shift left from last time, which would move those seats from pure ties to very slightly Liberal leaning seats.

Now, the rest of the board is mostly unchanged - the Liberals remain slight underdogs in Kildonan and Kelowna, ever so narrow favourites in a couple of outer Greater Vancouver seats, and Atlantic Canada is a quagmire of local ground reports versus polling. Right now I’ve got 7 seats going Conservative in the east, but it’s only 5 on the model (and possibly four, given how good the Atlantic polling has been today, if that continues). The battle of ground versus top level math continues.

What’s it all mean? Probably nothing, and yet everything. The Liberals remain on track to win. It is now nearly impossible for the model to show anything other than a Liberal majority. Most of the arguments against a Liberal majority boil down to either pollster error or collective modeller error, and the dirty little secret is everybody doing this on the right sounds suspiciously like the people pretending Bonnie Crombie had a chance in February. Now, the Liberals could project for 185 seats, get 165, and yes I’d look like an idiot, but the polls would have to be so catastrophically wrong to make a genuinely unclear result possible that it seems highly unlikely. And also, if it’s 165 Liberals, people claiming Poilievre would win will still be more wrong than me, which is nice.

Updates every 20 minutes or so.