The Washington PostDemocracy Dies in Darkness

The electoral outlook for Democrats shifts from bleak to blurry

A voter checks in Aug. 9 at Suffield Middle School in Suffield, Conn. (Jessica Hill/AP)

At its heart, it’s all anecdotes. Every component of trying to figure out what’s going to happen in an election is based on some individual data point that jostles against myriad others. Some are bad: I talked to a guy at a gas station and this is what he said. Some are useful: I polled 600 people in this time period and weighted their responses to match the population. But each is a thing observed in one place at one point in time that floats to the bottom of the teacup to be evaluated.

Okay, fine. Let’s accept that as stated. And now let’s eyeball those tea leaves — because the pattern they’re revealing is noticeably different from the one revealed two months ago.

Sign up for How To Read This Chart, a weekly data newsletter from Philip Bump

One of the interesting patterns shown in polling this summer has been the divergence between views of President Biden and his party’s position on the generic ballot. Or, in English: While Biden’s approval rating is remarkably low, which often correlates to poor performance in November, Democrats are holding their own with Republicans in polls that ask people which party they plan to support. Biden vs. expectations? Not doing great. Democrats vs. Republicans? Doing all right.

What I’m talking about here, by the way, is polling averages; specifically those compiled by FiveThirtyEight. Polls are images; averages are filmstrips. And since late June, two things have happened to the averages of polling on Biden’s approval and the generic ballot: The latter has gotten better for Democrats, and the former hasn’t gotten worse.

We can look specifically at the announcement of the decision in Dobbs v. Jackson Women’s Health Organization, the Supreme Court decision that overturned Roe v. Wade. Supporters of access to abortion pledged that the court’s action would trigger political blowback, a promise that is often made in politics but that doesn’t always manifest. Since that decision was announced on June 24, Biden’s net approval (the percentage who approve of his job performance minus those who don’t) went from minus-17 to minus-16, a subtle change. The generic ballot margin went from a two-point Republican advantage to about even.

It’s useful to compare that to the same period before the decision. In the 90 days before Dobbs, Biden’s net approval dropped more than six points and the generic ballot was flat. That makes the apparent inflection point around Dobbs more pronounced.

This doesn’t necessarily mean that Dobbs is the reason for the inflection, certainly. For example, gas prices peaked in mid-June; if one considers views of Biden as distinct from views of his party (as seems broadly warranted), views of his presidency may be driven more by things like gas prices and movement on policy issues than Dobbs, over which he manifestly exerts less influence.

Left unresolved is what this means. Historically, low presidential approval has meant poor performance by the president’s party in midterm elections. But it’s unusual in recent years that a president would be viewed very poorly even as his party is running even in the generic ballot.

FiveThirtyEight’s Nathaniel Rakich calculated polling averages for the past four midterms, allowing this point to be made numerically. Here’s where net presidential approval and the generic ballot stood 90 days before each and where they stand now.

Net approval
Party on generic ballot

One of those things, as they used to say when I was a kid, is not like the others.

We’re talking about five election cycles here, of course, and we’re talking about 90 days out. In past years, the generic ballot has shifted quite a bit in the final three months of an election cycle. In 2010 and 2018, the final margin in the polling average landed near the actual national House vote. In 2006 and 2014, it didn’t — in both cases favoring the Republican Party.

The 2014 example is particularly useful: 90 days out, Barack Obama was a bit more popular than Biden is now and Democrats were doing a bit worse on the generic ballot. Then the party’s position collapsed.

This polling is not the only thing to look at, though. In two recent congressional special elections, Democrats have outperformed their 2020 vote share across precincts — meaning that Democrats did better than Biden himself two years ago. Polling is fine, but polling can have difficulty measuring things like surges in turnout spurred by unpopular Supreme Court decisions. Actual elections capture that better.

(There was also the constitutional amendment vote in Kansas, of course. That it was explicitly about access to abortion, though, complicates things. It was rejected robustly in a red state, but it’s hard to disentangle its being a specific evaluation of the issue at the center of the political discussion.)

Where does that leave us? With an unsatisfying shift: What looked six months ago like a robust Republican rout this November is now something murkier. Will Democrats continue to improve on the generic ballot as Biden stays flat? That could help the party break what Biden’s approval rating suggests will be a rough year for his party. Or will the energy in the race be shifted back in the other direction? Republicans are pledging that the FBI search of Donald Trump’s Florida home will trigger political blowback, a promise that’s often made etc., etc.

Should 2022 move from a strong to weak Republican year, it could validate one controversial move by Democrats: boosting extreme Republican candidates in primaries with the expectation that they would be easier to beat in November. In a year that favors Republicans overwhelmingly, that’s fraught! In a year that’s closer, it may pay off.

Sifting through these anecdotes is interesting and revelatory about American politics, but its predictive power is shaky. We like to talk about what might happen, but, of course, we simply have to wait and see. Is that a cop-out? It’s not not a cop-out, certainly.