Pollster Wars And The 5 Stages Of Political Grief

On Nanos And Conservatives’ Delusions

Here’s the dirty little secret about Canadian polling right now, and basically every summer: the “movement” is as often about which pollsters have been in the field recently as it is about the trends within those pollsters.

It’s better now that Liaison is also out with a weekly tracker and Abacus is every other week, but there used to be periods where Nanos were the only game in town some weeks, and 3 of the last 4 or 4 of the last 6 polls had been from one pollster. It can mess with our collective perception of reality, thinking that things have either tightened or blown out because of the volume of noise about it.

You’ve seen it in the US much more clearly this election cycle - at the beginning of June Democrats were up 8 in RCP’s (flawed) average, a number which fell below 5 at one point. It’s now back at 7.2%, not because there has been a miracle Republican comeback and then a similar fall, but because the better Democratic pollsters’ results aged out of the average, were replaced by pollsters more favourable to Republicans, and then the first set of pollsters came back into the field. (For what it’s worth, in Nate Silver’s much better average, Democrats have gone from up 7% to up 6% to now up 6.6%, because he knows how to ensure this doesn’t happen.)

Right now, yes, you can make a very limited case that the CPC are recovering slightly. Angus Reid’s got a L+4 only, Abacus is a L+6, and Liaison is back to single digits this week. It looks a bit better, if you exclude Nanos, which currently has a fucking 19% lead. And in the context of 3 recent polls showing a smaller Liberal lead, you can talk yourself into recovery. But it’s bullshit.

Abacus has, for methodological reasons, a sample that doesn’t fluctuate very much and has a higher Conservative floor. (My guess is that it’s recalled vote - some number of actual 2025 Conservative voters remember voting for Carney, so a sample that’s 43% people who remember being 2025 Conservative voters is likely an electorate that in actuality voted for Poilievre more than Carney, in the same way Abacus’ recalled vote showed the NDP way too high in 2025.) Angus Reid showed bigger leads earlier this Parliament, but they’ve never been exactly the Liberals’ favourite pollster, and their poll before was a L+5. Those two aren’t proof of real tightening.

If you want a genuine recovery narrative, you’d need something very different - you’d need a pollster showing a double digit Liberal lead to show that lead tumbling. We know it won’t be Frank Graves, whose tweets have said EKOS’ big sample polling has the CPC on 27% these days (and had a few days previously said LPC at 42%). So if Leger or Pallas come through with some kind of big swing right, then we’ll talk. But they haven’t yet, so it’s still nonsense.

This kind of composition effect nonsense is just that, because it’s an effort to contort yourself into a position that is indefensible. I know, because I’ve done it before, many fucking times. And so, after a probably unnecessarily long intro, let’s go through the 5 Stages Of “Does My Party Have To Fire Its Leader” Polling Grief.

Denial

This is the easiest one - it was how every Conservative was in the two months before the 2025 election and how Liberals were in 2023, when they were using selected crosstabs of certain pollsters to claim everything was fine. This is usually accompanied by arrogant statements like “how could anybody think this trend is real, it’s obviously nonsense.”

Anger

This is where the pollsters and/or the media companies sponsoring the polls are getting yelled at. Today’s example is people being mad that CTV - the firm that paid Nanos to conduct their poll - wrote up an article about that poll and not the broader polling landscape. Other examples include the time Don Davies called Angus Reid a Liberal pollster, every time someone claims are biased for the politics of the commissioning source, and every attack on the modellers.

Bargaining

This is the most mentally fun stage, where you have on some level accepted that your party is in a bad position, but you have a lot of theories about how it could get better. This is where you seee a lot of When and Once statements - “When the campaign starts and the media focuses on our leader, we’ll surge,” “Once we start running ads,” and what not. They’re almost never true, but at least you’ve accepted the fact that the satiation as it currently applies is untenable.

Depression

What I like to call the “if we’re gonna lose anyways, why even bother?” stage. The three months after the Liberals lost the LaSalle byelection and before Chrystia Freeland resigned were the funniest time for this, when people would genuinely claim there was no difference between the Liberals losing with 100 seats and 40 seats. Even before we had any idea that Carney would be this effective, I supported a Leadership change based on the simple idea that losing less badly was a worthwhile investment. That was somehow a controversial idea amongst Liberals.

This is where most Conservatives seem to be. There are some still in the earlier stages, but right now it appears that the majority of Conservatives know that things aren’t working, but aren’t prepared to get to the logical conclusion. Every line of copium about “who would do better?” is firmly in this box.

Acceptance

It will be interesting to see how quick Conservatives get here, because objectively their polling is atrocious and the fundraising data released this weekend suggests that Poilievre’s greatest strength as a leader is now fading. The fact that Conservatives have already hit Depression in the 5 Stages Of Political Grief shows just how bad their plight is. The question is less whether Poilievre can win the next election, and honestly more whether he can survive the year. At some point, the pressure’ll mount, if only out of desperation.