Nanos, Midterm Polls, And High Class Bullshit
Get Off The Poller Coaster
I understand why CTV is acting like Nik Nanos’ latest tracking means something important and valuable – they’re one of his media sponsors, and therefore it is in their corporate self interest that we give a shit what his data has to say.
What I don’t get is why everyone else is acting like this matters.
The Liberals are below 30% in Nanos and down 7% for two reasons – first, that the NDP are at 21%, as the usual tradition of third parties polling better between elections happens, and because Nanos has the PPC at 2%. My understanding years ago was that the PPC didn’t get prompted on initial call by Nanos – you had to choose Others, before being prompted for the PPC. If I’m still right about this (which I think I am, mostly just because their PPC readings are substantially below Abacus, who have them at 5% and the last Leger had them at 4%), then there’s probably a point or two of voters just saying CPC who would, if given the PPC as a prompted party, would say PPC.
Does any of this matter? Right now, no, and if this were an election campaign and we were getting wildly different PPC readings between Nanos and the online panels, I would write a 4000 word essay about the Australian Greens in 2007 and One Nation and Palmer now and the impact of prompting for smaller parties on polled vote intentions, and you’d be bored to death but use it as an explainer on why their vote won’t be whatever the various polls say.
Abacus this week shows essentially no change from 2021, but shock of all shocks, the media is running with Nanos, because of course they are. Plainly, if you believe a poll that has the Tories up 13% in Ontario, you’re letting yourself be lied to in the name of data that you desperately want to be true. You know how I know?
I genuinely believed Doug Ford would lose.
I don’t bring up Ontario 2022 that much, for obvious reasons – I’m embarrassed by the way I conducted my coverage of that campaign, and how I made a huge mistake. I’ve written before that the logic that got me to Steven Del Duca governing after that election was 2 parts – that the NDP would collapse back (ding ding ding!), and that because the PCs had a ceiling in the mid-to-high 30s, the Liberals would (inevitably) rise to fill the void and win essentially by default. Obviously, that second part was dead, dead wrong.
The problem with a lot of my analysis at the time is that it suffered from what could be called “high class bullshit”, in that there was a data backed case for everything I said, even if what I was doing wasn’t really in the whole based in the data. Hilariously, given that Ontario causes me much less external reputational damage than 2020 US does, the Ontario coverage was worse, because the US was really just that the polls were dogshit. In Ontario, the polls consistently showed a story of Tory control with a lane for the Liberals – a lane they never managed to use or take advantage of – but I painted that lane as the likeliest, if not only, path left.
The problem with Nanos isn’t even so much about the poll – I mean, yes, the way they do their insane, 4 week, 1000 total person sample means that it is needlessly slow to react and of limited value compared to others – but more about the coverage. Who gives a fuck about whether a Prime Minister is losing by 4 or 7 years out from the likeliest next election?
If you think Justin Trudeau should be leading in the polls right now you have a shakier grasp on reality than you should. Inflation is high, interest rate hikes are squeezing homeowners, and we are all collectively still feeling the effects of a pandemic which, let’s be honest here, we have not properly reckoned with the toll of. Be it the grief of lost love ones, the regrets of missed moments, or the tragedy of isolation, the pandemic has our people, and our country, on edge. Expecting everyone to love their government right now is a bit much.
But again, I keep coming back to Ontario, because there was a lot of polling showing Ford’s PCs in the fucking shits at various times midterm, and even closer to the campaign. Not all of the polls said it, but plenty of them did. Look what happened. Polls at midterm of a government are not often useful as a predictive metric, because polls at midterm tend to be referendums on the incumbents. This was true in 2020 and 2021 when the Liberals had huge leads, and it’s true now. When the 2021 election was called, people started judging their leaders more harshly, and voters who weren’t traditional Liberals but were fine with the government went home to the CPC.
Did Christy Clark and Kathleen Wynne and Dalton McGuinty and McGuinty again not prove the point about polls during a government’s tenure being about as predictive as I am straight at predicting eventual election results? Didn’t Pauline Marois have a huge polling lead for most of 2011 and 2012 before winning 4 more seats than the Liberals in 2012? (Update: no, but that was because the CAQ led a bunch of polls in 2011, which, still doesn’t disprove my point.)
Could this be written off as a bunch of whining from someone who wants the Liberals to win? I mean, sure, but if you genuinely think that I am willing to in perpetuity go to bat for the left against all sense, look at the way I cannot talk myself into Rachel Notley even when the math says she’s at worst a coinflip to win on current polls. My contention is that when the rubber hits the road, the Liberals will win. That doesn’t mean they will be winning all the way along the road.
This site would be immeasurably better off – in terms of traffic, readership, and notoriety – if I wrote a response column to every Federal vote intention poll. Just, objectively, using every poll release as a way to grind traffic, to grind Twitter interactions, and to drive attention to myself as a short term strategy would work. I know this, and I can’t do it, because fundamentally I do not believe it matters. I will not believe it matters if in 6 months the Liberals take a lead either, just like I am not using Mississauga Lakeshore as evidence that the polls are wrong, which is 1000% what Scrimshaw of 18 months ago would be doing.
I believe when the rubber meets the road, the NDP will get 15% of the vote and the Liberals will not suffer greater consequence in 2025 from left wing vote splitting than they did in 2021 or 2019. I believe when Poilievre actually has to take positions on legislation, and has to decide whether to continue to fund the child care deals and the Dental and the Pharma, he will face a choice – pissing on the benefits accrued to those people and staying true to his base, or pivoting and pissing them off. If he pivots the same way that O’Toole and Scheer did, then he’s inauthentic and a fraud. If he doesn’t, how does he win the suburbs.
All of this might be wrong – I don’t think it is, because of course I don’t, but absolutely none of what I am actually predicting, and none of what should actually make up a case for Poilievre skepticism, is proven or disproven by Nanos’ weekly readings. As the guy who has taken victory laps about electoral predictions on the basis of polled vote intentions that ended up being dogshit before, trust me on this one. Nothing happening in the polls right now matters fuck all for the next election. It might be fun to watch anyways, but it’s fun in the same way me playing video games is. It’s not important, it’s just fun.
Just don’t pretend it actually matters.
Nanos says that it refers to the PPC by name: https://nanos.co/dataportal/nanos-tracking-methodology/
Important to note that Abacus uses a non-probability sample, while Nanos uses a probability sample, generally considered superior. That aside, I agree that polls years before an election are just to fill media time/space and as valuable as predicting the Dow Jones in a year or two.
Rereading this again. Given The Star is now getting Abacus to churn out "exclusive" polls for them every week, it doesn't look like your imploring the media to get off the rollercoaster has worked. Not surprising. It gives them content to write about, fits in with the preferred 30 plus year old narrative that Canadians are now 'tired' of a three time elected government and is likely cheaper than hiring actual journalists. Sigh.