Ohio, PA, State Polls, And Wanting To Believe In Lies

State Polls, Boy, I Don't Know

One of the great debates of my life is what matters more, process or results. It’s a bit of an internal war for me, as the nerd in me prioritizes process and the sports fan in me only cares about process so much when the results – as unsustainable as they may be – is driving me crazy, either in good or ill. It’s one that’s especially on my mind as I think about my F1 fandom cheering for Mr. Consistency, George Russell, who is the only driver in the grid to have 10 Top 5 finishes this season.

Russell has been genuinely impressive often, but othertimes quite lucky to keep that streak alive, and that dichotomy wracks me as I try and figure out what will happen when Mercedes stops giving him a car unable to compete with the front. It’s this hard concept, and the more and more George puts up the results any way possible, the less I care about the process and the more I just accept that George is a future World Champion.

And I’m thinking of all of this as we get more and more Ohio and Pennsylvania polls that point to just a truly unbelievable result for Democrats.

…

If you take the 538 average – which adjust for partisan bias of the polls – Tim Ryan is up 1.8% in Ohio. If you take a straight average of the three listed 2022 Pennsylvania polls, John Fetterman is up 6.3%. If you adjust the fact that one of those polls is a GOP poll, your average moves to Fetterman +7%. Fetterman up a bunch and Ryan up a bit passes one of my tests for a set of polls, which is that they’re internally consistent – if Ryan is winning, Fetterman is up by a bunch. By the other test – plausibility – they’re a fucking disgrace.

Ohio voted for Trump by 8% in 2020, while Pennsylvania voted for Biden by 1%, so it’s perfectly plausible that these polls, in aggregate, would have a delta between these two races in the high single digits. It’s even plausible that candidate issues in Ohio would have Ohio trending left relative to both the nation and Pennsylvania, with the combination of JD Vance being shit at this and Tim Ryan having a very good history of outperforming with white working class voters. But what’s happening now is that the volume of good data points for Democrats is making people go “at some point you have to ask what’s going on here”. Spoiler: you don’t, actually.

I could maybe buy the idea that Dr. Oz is such a bad candidate right now that an election held today would have Oz down 4-6%. I could maybe get there, but if I did, I would do so by acknowledging that the likelihood is that the undecideds are Republicans waiting to come home. But Ohio? Nah, these polls are just wrong, and the fact that the Ohio polls are so wrong makes me think that it’s just the same polling problems again.

Caring only about outcomes and not processes makes believing things like “Ohio Senate is a tossup” easy, and in some ways, I wish I could believe it. I wish I could see these polls and think “yes, let’s go!” and not “what drugs are these pollsters on?”, because it would be a lot more fun! But I know it’s bullshit and I know that these arguments are just purely wastes of time because we have nothing else to argue about, so far away from the elections themselves.

If you modelled Ohio Senate with neutral candidates and a tied national environment – an estimate that is probably too kind to Democrats – you’d have the GOP up notionally somewhere between 12% and 15%, depending on whether you use a trends model or a straight swing model. To get this race to within 5%, you need to either believe that it will be a Democratic-leaning national environment (probably not) or that the gap between Tim Ryan and JD Vance is worth 7-10% above generic Dem and generic GOP. To think he could win, you’d need to think that gap is worth up to 15%. Outside of Jon Tester, Amy Klobuchar, Joe Manchin, and Susan Collins, nobody in the current Senate could be considered as a plausible candidate of that talent. It’s possible Ryan is that good, but if he actually believed that, he would have run for Governor in 2018 instead of waiting to run statewide when his seat was abolished.

In Pennsylvania, that same model would have the GOP up ~6% before candidate quality, meaning to believe the current polls you’d need to think Fetterman and Oz would be worth about 12% statewide compared to Generic Dems and Republicans. Does that sound likely, given the nature of increased polarization and increased Presidential partisanship dictating Senate outcomes? No?

Put it another way, would you believe a poll that said the GOP were winning Rhode Island’s 2nd district? Hell, did you believe that Suffolk poll that had that? I didn’t, because it doesn’t pass a basic smell test – the GOP nominee is well known, the Democrats weren’t, the undecideds probably leaned left and Democrats will be fine. Believing that Democrats can outperform by such distances because of the polls but not believing Republicans could outperform by the same margin is absurdist hypocricy of the kind I was guilty of in the past.

Remember the 2020 Arkansas June poll that was Trump +2? I wrote then that if this swing was even half right, Texas was gone and it would be a landslide. Turns out, Joe Biden got about the same swing in Arkansas he did nationally – around 2%. The idea that “even if these polls are numerically wrong they’re directionally correct” isn’t even correct in the US these days – they’re just out to lunch and making the same mistakes they’ve made for years.

Look through the column archives of this Substack – most of my early work is uncut Democratic optimism about 2022. These state polls would validate that archive and allow me to say that I was right the whole time. I would like to believe them. I would love to believe them, in the same way that I am becoming increasingly of the view that George Russell is going to win the World Driver’s Championship in the next 3 seasons. But in my core, I know they’re wrong, and all you’re doing by believing them is lying to yourself and making November 8th so much harder for yourself than you need.