US House: Generic Ballots And Guessing Metrics

The GOP Started Something They Couldn't Finish

What is the current Generic Ballot?

It’s a stupidly hard question to answer, mostly because there are about 7000 different ways you can slice the data to tell you any story you want to tell at this point. If you want to choose extreme optimism for Democrats, you’re a huge fan of real things, like election results in the five post-Dobbs special elections, and if you’re Republicans you care a lot about primary turnout. If you’re attempting to be neutral, generic ballot polls are in the mix, but there’s also the questions about state polls and Biden’s approval rating in the mix. Oh, and there’s the little matter of the 3 Gubernatorial elections last year too, which of course were (in aggregate) amazing for Republicans.

Everyone has their preferred metrics or even their preferred mix of metrics – I don’t think primary turnout is at all useful, and I don’t think state polls are either, so at least the two metrics I hate mostly cancel each other out – but what if you just combine the metrics? What would that look like?

If you take the 2020 to 2022 swing in the 15 special elections this term, plus the 2020 to 2022 swings in the 3 Gubernatorials, throw in 2018 to 2022 swing in the Washington and California Top 2 primaries, the 538 Generic Ballot polling averages, and then a pair of my own metrics around Biden’s approval and the Generic Ballot (designed to correct for the pollsters’ inability to poll minorities), you get a wide mix of metrics that all point in different directions and certainly point to different outcomes.

All in, that mix (with the 5 post-Dobbs specials given double weight over the 10 pre-Dobbs specials) gets you a D+1.1 environment – hilariously close to the current 538 average of a D+0.9 advantage right now. Does that make it correct? I have no idea – this is an explicitly untested metric, and proximity to another metric is not in and of itself a sign that this is successful – but it’s notable, at least.

In the same breath, though, this metric does have the advantage of being something that both Democratic optimists and Republican Red Wavers can hate in it, which probably means I’ve done something right. The Democratic optimists will point out that Virginia and New Jersey were held at an entirely different time and in an entirely different environment, and therefore those results are meaningless to predicting elections post-Dobbs. On the other side, using the special elections as a metric – and giving double weight to the post-Dobbs ones – will give ire to those who believe they are useless metrics, a take I don’t even necessarily disagree with. It’s certainly the case that I don’t think it’ll be a D+10 environment, which is what the 5 post-Dobbs specials would point to.

In this case, the average of the metrics is a bit of a defence mechanism – we have no idea what matters this year, so by adding in so much to the model, the chances of any individual metric fucking us by being wildly wrong falls apart. In this case, I doubt that Virginia and New Jersey mean much if anything in a post-Dobbs world, but I’m also not stupid enough to realize that I want that to be true – and that if this was a GOP President and we had gotten those sorts of swings in those states, I’d be trumpeting them forever, so they go in the metric.

So, if it’s currently a D+1.1 – and while this metric isn’t going to be too volatile, with the fact that 50% of the inputs are things that have already happened and therefore are locked in (2021, Specials, and the Top 2 Primaries), it still is sensitive to continued rises or falls in the polls – what does any of it mean?

Everything and nothing.

…

The “everything” is that a D+1.1 Environment would likely elect a Democratic House, given cash advantages, incumbency, and the fact that the GOP swing looks to be inefficient – bigger swings in heavily white working class seats and smaller swings in more suburban ones ends up delivering less seats per point of national swing than the GOP would want. That is not to say if the polls on November 8th say a D+1.1 the House is Likely Democratic – it wouldn’t. because of risk of polling error. But if we locked in the November House popular vote at D+1.1 (adjusted for uncontested races), I’d bet the Democrats have won the House.

They won’t have won it hugely, but at a D+1.1 Democrats are definitely winning WA-08, MN-02, VA-07, and KS-03 – the tier of left-trending suburban seats that the GOP need to win for a ~230 House majority – and are probably losing VA-02, CO-08, MI-07, and NY-19, left-trending seats that aren’t as blue and/or don’t have Democratic incumbents. If it’s a D+1.1, Cindy Axne in IA-03 is in play to win again, and the two narrowly Biden Arizona seats – AZ-01 and AZ-06 – are at least somewhere on the board.

A D+1.1 would put Democrats in with a solid chance of 5 seats in Ohio, even if Kaptur in OH-09 and Sykes in OH-13 would be in pure coinflip races, and 8 seats in Michigan. In such a scenario, the GOP should give up on MI-03 and focus on their efforts on John James’ quixotic efforts to carpetbag into the 10th, in all honesty. In Pennsylvania, a D+1.1 gives all three of Cartwright, Wild, and whoever the fuck we’re running in Lamb’s old seat (I cannot be assed to remember it or look it up) a chance of winning, which is a minor miracle.

The ”nothing” of this is the fact that this metric saying this on September 2nd means fucking nothing for what will happen in 9 and a half weeks, when the voting actually happens. It is beyond moronic to suggest that there is any meaningful way of assessing the environment in November, because all the usual metrics (namely, gap from historical polling to result) is based on extremely small samples, and also would have suggested that Democrats should be fading, not gaining, ground right now. More bluntly, Dobbs cannot be understood as a radical departure from the historical norm while then plugging in a historical poll movement metric, because it either is a history defining event or it isn’t.

Are Democrats favoured to win the House right now? My instinct is no, but that’s not really based on anything except the fact that Democrats usually disappoint me, but if Dobbs is truly a history defying event, then there’s no real case for the House to not be a tossup. The only case for the House to not be one is that the GOP will do better in November than they’re doing now, which is probably 50/50 without a historical prior about what midterms usually do. That said, I’m not really letting myself believe too much that a Democratic House is more than a 40% chance.

The inflection points from here don’t suggest a GOP resurgence either – Schumer is teeing up votes this month on insulin and gay marriage in the Senate, and if they pass, then Democrats get to add one or more legislative accomplishments to the list, and if they don’t, the GOP lose yet more news cycles on how evil they are for opposing equal rights and cheap medicine for the ill. The January 6th Commission will probably have more hearings, which is more oxygen out of the system, gas prices are trending down (but a bad hurricane season could fuck that up), there’s a couple more inflation reports which would be key … but the GOP are going to be facing a news cycle that they increasingly are responding to, and not setting.

When I walked into the liquor store I hit up yesterday, the Smiths’ This Charming Man came on, my favourite song from an artist that anybody who’s read Salvation In The Storm knows I love so much, and since then, I’ve been listening to them a ton. And as I think about all the contradictory ways to analyse the House at this point, all I can do is think of them. “I started something, I forced you to a zone/And you were clearly never meant to go” could be the best descriptor of the position the Supreme Court put the Congressional GOP in, because the GOP wanted to be able to ride on discontent and bad vibes to 235 seats in the House and 53 Senate seats, and now they can’t.

What’s gonna happen in the House? Honestly, I might just have to quote them again – I’m not too sure.

And frankly, nobody else should be either.