Notes on the midterm elections

Saturday night, when Nevada had counted enough votes to declare Catherine Cortez Masto re-elected, we learned that the Democrats would hold onto the Senate. They may even gain a seat, if Raphael Warnock can win the December 6 run-off in Georgia.

As of this morning, almost a week after election day, Republicans are leading in the House, but still have not nailed down a majority. 212 races have been called for Republican candidates, 204 for Democrats. 218 are needed for a majority. NBC is estimating that when all the counting is complete, the GOP will have a slim 219-216 majority. (So assuming Lauren Boebert hangs on to her current slim lead, Speaker McCarthy will lose any vote in which he can’t get Boebert, Matt Gaetz, and Marjorie Taylor Green’s support.)

In the states, Florida went very red, but both Michigan and Minnesota very blue. Democrats flipped governorships in Massachusetts and Maryland, Republicans in Nevada. Arizona is still undecided.

There’s probably a lot to learn from these results that I haven’t deciphered yet. But here are a few conclusions that seem obvious.

Voters in swing states don’t believe the Big Lie about 2020, and want to continue having democratic elections. My biggest fear about the midterms was that they would herald the end of democratic elections in the United States. But that didn’t happen. Yesterday, the NYT’s home page included the headline “Every election denier who sought to become the top election official in a critical battleground state lost at the polls“.

During the 2020 election, it was secretaries of state — both Democrats and Republicans — who stood up to efforts by Mr. Trump and his allies to overturn the results. State election officials certified vote tallies over Republican objections, protected election workers from aggressive partisan poll watchers and, in at least one case, refused a personal entreaty from the president.

The next spring, several candidates pushing the false narrative that the 2020 election had been stolen announced their intention to run to be the top election officials in critical states.

Republican candidates for secretary of state in places like Arizona, Nevada, Colorado, and New Mexico claimed the 2020 election had been stolen from Donald Trump, without basing that belief on the slightest bit of evidence. In Pennsylvania the secretary of state is appointed by the governor, and gubernatorial candidate Doug Mastriano was running as a rabid election denier (in addition to being a Christian nationalist and a barely-in-the-closet antisemite). In Georgia, incumbent Republican Brad Raffensperger had beaten back an election-denier challenge in the primary.

Victories by election-denying candidates would have opened the possibility that in 2024, MAGA secretaries of state might refuse to recognize a Democrat’s victory. If they needed no evidence beyond Trump’s say-so to declare fraud in 2020, why wouldn’t they do the same in 2024?

Fortunately, all those candidates lost. Some of the races were disturbingly close — though Mastriano got soundly thrashed — but they lost. Only Indiana chose an election-denying secretary of state. That could be a problem locally, but it’s unlikely to affect a national election, since a Democratic presidential candidate could only carry Indiana in a national landslide. (Barack Obama barely did in 2008.)

Sweeping abortion bans are unpopular. A lot of Americans have conflicted views about abortion, so the wording of a proposal matters. But if you put a broad abortion ban in front of the voters, they’ll reject it even in some pretty conservative states.

We saw that already in August, when Kansas (which Trump carried 56%-42% in 2020) held a referendum that would have given the legislature the power to ban abortion. (The state’s supreme court had found a Roe-like right to abortion in the state’s constitution. This proposed amendment would have removed that right.) The legislature scheduled the vote to coincide with a primary election that was likely to draw more Republicans than Democrats, but it didn’t matter: Turnout was huge and the proposal failed 59%-41%.

Tuesday, proposals to protect reproductive rights were on the ballot in Vermont, California, and Michigan, while voters had a chance to restrict abortion rights in Kentucky and Montana. The pro-choice side won all five.

The abortion issue is also getting credit for the Democrats avoiding the typical midterm-election-collapse of a party in power. It’s hard to say precisely why voters decide to show up and lean one way or another, but the turnout of young voters was high and heavily Democratic, and Democrats won 68%-31% among single women. Chances are that abortion had something to do with that.

[BTW, it has been hilarious to watch conservative pundits struggle not to grasp that single women don’t want the Republic of Gilead controlling their bodies and making major life decisions for them. Fox News’ Jesse Watters noted that married women tend to vote Republican, so he had a solution: “We need these ladies to get married. And it’s time to fall in love and just settle down. Guys, go put a ring on it.” One America News’ Addison Smith went even further down the patriarchal rabbit hole: “Secular progressivism has turned people into masochists. … 68% of single women voted for people who vowed to let them legally murder their children and continue on living miserable single lives without purpose, without responsibility or meaning.” Attention single women: After a quick search, I wasn’t able to determine whether Addison Smith is married, so he might still be available to bring purpose and meaning to your otherwise pointless life.]

Trump is hurting the Republican Party. He wasn’t losing any races himself this time, but he screwed up the GOP in two other ways. First, the unqualified and too extreme candidates he pushed to victory in the Republican primaries went on to lose winnable elections.

New Hampshire is a good example. Republican Governor Chris Sununu, who kept his distance from Trump, cruised to a 57%-42% victory. But Trump-endorsed election-denying Don Bolduc lost to incumbent Senator Maggie Hassan 54%-44%. Those two results are from the same voters on the same day, so about 1 NH voter in 8 must have voted a Sununu/Hassan split ticket.

Similarly in Georgia, Brian Kemp (who defeated a Trumpist challenger in the GOP primary after certifying Biden’s 2020 victory) won the governorship 53%-46%, while Trump’s handpicked senate candidate Herschel Walker faces a run-off after trailing 49%-48% Tuesday.

The second way Trump undercut the GOP was to divert attention towards himself, his petty grievances, and his backward-looking complaints about 2020, and away from issues like inflation that were working for Republicans. He doesn’t seem to realize that if the 2020 election were run again, he would get his butt kicked again. Most unpopular presidents see their images improve in hindsight, but not Trump. 54% of the public still views him unfavorably.

Conservative power brokers like Rupert Murdoch and mainstream Republican politicians like Paul Ryan see what’s going on and would like to free the party from Trump’s destructive influence, but I’m betting against them. There’s a lot of Trump-blaming in conservative media right now, but that just means it’s January 7 again. Before long, the would-be rebels will be crawling to Mar-a-Lago to kiss the ring, as they did only weeks after January 6. The NYT’s Jamelle Bouie agrees with me:

The idea that Republican elites could simply swap Trump for another candidate without incurring any serious damage rests on two assumptions: First, that Trump’s supporters are more committed to the Republican Party than they are to him, and second, that Trump himself will give up the fight if he isn’t able to win the party’s nomination.

I think these assumptions show a fundamental misunderstanding of the world Republican elites brought into being when they finally bent the knee to Trump in the summer and fall of 2016.

But Jonathan Chait disagrees. It’s different this time, he says, because Ron DeSantis provides a real alternative.

Either way, it’s going to be ugly. Trump has already shown that he will try to burn down a democracy that won’t re-elect him. No one should be surprised if he burns down a party that won’t re-nominate him.

Gerrymandering matters. For years, voters in Michigan have voted for Democratic candidates for the legislature, only to see Republicans keep control. But in 2018, Michigan voters reestablished democracy in their state by overwhelmingly passing Proposal 2, which created a nonpartisan redistricting commission. Tuesday was the first election held under the new nonpartisan maps, and Democrats won majorities in both houses for the first time in almost 40 years.

Compare Michigan to Wisconsin, which is still heavily gerrymandered in Republicans’ favor. Democratic Governor Tony Evers was re-elected with 51% of the vote, and Republican Senator Ron Johnson was re-elected with just over 50%, suggesting an evenly divided electorate. But Democrats narrowly avoided a veto-proof Republican supermajority in both houses of the legislature, which would have made Governor Evers virtually powerless.

Both parties gerrymander when they can, because it’s political suicide to let the other side play by different rules. (Though no Democratic-controlled state mirrors Michigan, with an entrenched Democratic legislature thwarting a Republican majority in the electorate.) But Democrats want to end this game: An anti-gerrymandering provision was part of the For the People Act, which has passed the House in 2019, 2020, and 2021, only to be blocked by Republican filibusters in the Senate. A scaled-down proposal, the Freedom to Vote Act, was put together by Senators Manchin and Klobuchar. But it also was blocked by a filibuster.

The stuff Biden did is way more popular than Biden himself. The pundits predicting a red wave were fooled by Biden’s low approval rating: 41.5% in the latest 538 average, barely higher than Trump’s 39.9%. Normally, a president with numbers like that sees his party get clobbered in the midterm elections.

But Friday, Chris Hayes made an interesting comparison to 2010, when there really was a red wave. In 2010, the Democratic Congress had just passed ObamaCare, and it was very unpopular. (Since it hadn’t taken effect yet, Republicans could claim anything they wanted about it, and they did.) The way Republicans ran against Democratic incumbents that year was simply to point to that vote.

Nothing in this cycle played that same role of connecting Biden’s unpopularity to specific votes in Congress. If Democrats got criticized for voting for the Inflation Reduction Act, they could say, “Yes, I lowered your prescription drug costs, invested in renewable energy, and created jobs for American workers.” The bipartisan infrastructure bill? “Yes, I voted to rebuild America’s roads and bridges, bring broadband internet to rural areas, and replace lead water pipes that have been poisoning our children.” American Rescue Plan? “I voted to get Americans vaccinated, send money to people who couldn’t work during the pandemic, and give loans to businesses so they wouldn’t have to fire people.”

And so on.

So sure, Americans are frustrated with inflation, and Republicans were able to fan people’s fears about rising crime and a few other issues. But how could challengers pin those problems on the incumbent senators or representatives they were trying to replace? And while 2010 Republicans could promise to repeal ObamaCare, what exactly were 2022 Republicans proposing to do about inflation, crime, or anything else?

Polls are more-or-less accurate if you don’t expect too much out of them. I’ve seen a lot of the-polls-were-wrong-again punditry, but I don’t think it’s deserved.

Take the Georgia Senate race, for example. 538’s final pre-election analysis said that Herschel Walker had a 63% chance of winning, with a predicted margin of 1.2%. But now that the votes have been counted, Warnock holds a .9% lead. (Warnock is still short of 50%, so a run-off is happening December 6.)

So Nate Silver’s prediction was off by 2.1%. You can’t really expect pre-election polling to be more accurate than that.

At best, a poll is a snapshot of where the electorate was a day or two before the election, accurate to within some margin of error. Averaging a bunch of polls (as 538 does) should shrink that margin, but not to zero. And as for people who decide at the last minute to vote (or not vote), or who change their minds in the booth — there’s really no accounting for them.

In short, the right response to the Georgia outcome is not “538 was wrong to say Walker would get more votes, because Warnock did”, but “538 said the race was going to be close and it was.”

That’s why the model said “63% chance”. Just for reference, NBA star Giannis Antetokounmpo is currently hitting 65% of his free throws (well below the league average of around 78%). That’s better than a coin flip, but when he steps to the line, Bucks fans are holding their breath rather than counting the points.

Is a more accurate system possible? Well, maybe, but you won’t like it. I suspect that somewhere in the basement of Meta headquarters, somebody has developed an algorithm that predicts how each of Facebook’s tens of millions of users will vote. (For most of us, it wouldn’t be that hard.) Facebook users may not be absolutely typical of the electorate, but the differences are probably not difficult to model and compensate for after you review the data from a few election cycles. And on election day, if the app on your phone is tracking your location, it knows whether you went to the polls.

Facebook’s huge sample base would eliminate nearly all the statistical error. Since it’s a spying algorithm rather than a request for information, you couldn’t just refuse to answer, eliminating another source of polling error. And in order to lie to the algorithm, you’d have to change your whole online behavior, which hardly anybody is going to do. The algorithm’s estimates would always be up to the minute, and in the end it would know a lot about who voted.

I’ll bet that system could be pretty accurate.

Expert speculation, on the other hand, isn’t worth the attention it gets. All that talk of a “red wave” didn’t come from the polls. 538’s generic-ballot polling average finished with a Republican advantage of 1.2%, which would lead a person to expect a Republican Congress, but not a sweeping rejection of Nancy Pelosi’s Democrats. (Compare that margin to true wave elections: In 2010, Republican House candidates got 6.8% more votes than Democratic candidates. In 2018, Democrats did 8.6% better than Republicans.)

The red-wave speculation came from pundits, both too-optimistic Republicans and too-pessimistic Democrats.

The right lesson to draw is that we spend way too much time listening to people speculate about stuff they don’t really know. Psychologically, it’s understandable: We get anxious leading up to an election or some other big event, so we want to believe that someone can tell us what’s going to happen. Even hearing that things are going to go badly can be more comforting than facing life’s real uncertainty.

It’s also understandable from the networks’ point of view: Actual reporting is hard and can be expensive, but gathering a panel of talking heads in the studio is easy and cheap. (A lot of them have a book to sell, a candidate to push, or some other reason they want to be on your show. So you may not have to pay them at all.) By air time, an investigative reporter may or may not have cracked whatever story she/he/they has been working on, but a pundit can be guaranteed to have a speculation ready on demand.

Unfortunately, those speculations aren’t worth much. If listening to them makes you feel better, fine. But don’t kid yourself that you’re receiving valuable information. Life really is terrifyingly uncertain.

So in the end, I wind up agreeing with the conclusion of the editorial I linked to about the polls being wrong again, if you change “study polls” to “try to prognosticate”:

Voters would do well to study the issues more than they [try to prognosticate], and media would do well to provide valuable issue-oriented reporting instead of reporting on a horse race that can change minute to minute.

[BTW: If political predictions were intended to be accurate, networks would keep detailed statistics on which pundits were right or wrong, and there would be bidding wars over the ones with the best records. That doesn’t happen, does it? The red-wave predictors aren’t going to lose their jobs, and somebody who got it right isn’t going to suddenly vault to the top of the profession the way a market-beating hedge-fund manager would.]

Post a comment or leave a trackback: Trackback URL.


  • Nancy Banks  On November 14, 2022 at 10:00 am

    Somewhat unrelated – but now that the Dems have lost Fl, can we finally lift the sanctions against Cuba. Our actions simply make no sense and only work to impoverish the country. Great goal for the USA.

    • weeklysift  On November 16, 2022 at 7:14 am

      Back in the Eisenhower/Kennedy years, the embargo was supposed to produce economic distress that would quickly topple the new Castro regime. That was a more-or-less reasonable policy at the time, but should have been re-evaluated when it clearly wasn’t working. Now it’s just ridiculous.

  • dmichael  On November 14, 2022 at 11:00 am

    “Polls are more-or-less accurate if you don’t expect too much out of them.” That compels the question: “Then what purpose do they serve?” Certainly not predictive value. The example you use shows that 538 got it WAY wrong, with Walker not winning by 1.2% (a significant margin) but losing, requiring a run off. Your sports metaphor is inapt. Free-throwing shooting involves an activity where the shooter has control over almost all of the variables and can practice exactly the same activity.
    Then why are there so many pre-election polls? Money and politics. In the major media environment with their “horse-race” analysis designed to get clicks, advertising revenue and subscriptions, the results are heightened anxiety and tension, just the things you (and I) try to avoid in this season. Many of the pre-election polls were conducted by Republican-leaning entities and were intentionally designed to depress Democratic voting.
    In one way, polls are worse than the blathering of pundits: They have the veneer of being accurate.

    • Dale Moses  On November 14, 2022 at 7:13 pm

      His example is fine. The problem is that he selected it. When you select an example rather than using the entirety of the construction you’re liable to introduce bias.

      The polling was bad not because this one was got wrong (it wasn’t) but because Nate’s methodology is bad and bad in a way that he should know and this caused him to get the general thrust of the answers very wrong. Specifically the issue is that he does not require methodology before accepting a polling outfit. But a polling outfit is not orthogonal to your model if they’re not operating in good faith. And as a result the polls are able to be moved by people who want to generate a narrative.

    • George Washington, Jr.  On November 15, 2022 at 7:05 am

      FiveThirtyEight works in percentages, like the weatherman. is an aggregator that makes specific predictions based on polling averages, and they were spot-on except for Nevada, although in a close race, that’s expected. They predicted Georgia would be a tie. They also throw out biased polls like Rasmussen and Trafalgar that could skew the average.

  • Brian Douglas  On November 14, 2022 at 11:05 am

    Slight disagreement. Nate Silver of 538 fame was primarily an evaluator of talent in professional baseball using analytics. It was only after the 2008 election (when Silver correctly predicted the outcome in 49 out of 50 states) that he rose to national prominence. So maybe being correct, does make one’s career. And I’m not even going to bring up Jean Dixon.

    • weeklysift  On November 16, 2022 at 7:25 am

      538 is an exception that points out a hole in my rule. It rose to prominence largely by going around the traditional media. Political junkies were following 538, and eventually mainstream media followed. And that happened during an era when blogs like TPM could rise to prominence and then get incorporated into mainstream media.

      What I would love to see is a larger pattern of pundits losing their jobs when they’re egregiously wrong for long stretches of time, and being replaced people who have been getting things right. Instead, I see a lot of talking heads like Larry Kudlow, who has basically NEVER been right, but is still on TV.

  • Daughter Number Three  On November 14, 2022 at 11:39 am

    I think you’ve got a typo in this paragraph:

    “Both parties gerrymander when they can, because it’s political suicide to let the other side play by different rules. (Though no Democratic-controlled state mirrors Michigan, with an entrenched Democratic legislature thwarting a Republican majority in the electorate.)”

    Shouldn’t that be *Wisconsin*, not Michigan?

    • weeklysift  On November 16, 2022 at 6:59 am

      Until the new legislature takes office, Michigan is still an example. In Wisconsin, there’s not a clear Democratic majority in the electorate.

  • Azetheros  On November 14, 2022 at 2:33 pm

    Two questions:

    1) A fair number of folks on the left would add to your observations that this was a landmark election for youth turnout, and a landmark election for progressive candidates (Fetterman, Lee, Flowers, Casar, and others I am forgetting). Do you think this is evidence that the Democrats should consider further progressive policy an investment in that electorate? Do you think it is categorically too early to tell? (And if so, do you think it is worth going out on a limb, or sticking to the current strategy?)

    2) Likewise, a lot of folks on the left (myself included, insofar as I have any kind of perch) are making a point of targeting Sean Patrick Maloney (outgoing head of the DCCC; personally bigfooted Mondaire Jones out of his district), disgraced former-Governor Andrew Cuomo, and Mayor Eric Adams for those losses. Cuomo for making the deals and appointments that led to New York’s redistricting debacle; Adams for legitimating the “crime!” message a lot of Republicans relied on, and; Maloney both for his own incompetence in his own race and for refusing to invest in toss-ups where the winner might be a progressive, like Jamie McLeod-Skinner in Oregon. (I’ll note, a LOT of the voices I am hearing on this are based in New York.) Any thoughts on whether this means, or should mean, and significant changes in the Democratic leadership?

    • weeklysift  On November 16, 2022 at 7:07 am

      The progressive/moderate question is complicated, and I still need to think about it. Fetterman is an interesting case — and let’s assume he would have won by more if not for the stroke. He has a mix of progressive and moderate positions (fracking, for example), and appeals to white working class voters with his image and manner of speaking. I think there’s something to learn from him about how to talk about progressive issues. But I don’t think it’s as simple as progressive or moderate being the way to go nationally.

    • weeklysift  On November 16, 2022 at 7:09 am

      I don’t know what happened in New York. I went into the election thinking of it as a secure liberal bastion, but for a few districts. So I didn’t pay much attention to it. That clearly was a mistake.

      • George Washington, Jr.  On November 16, 2022 at 8:01 am

        The Democratic-majority New York legislature attempted to gerrymander the heck out of the state, but the conservative judges Cuomo appointed threw out their map and replaced it with one more favorable to Republicans. The same process happened in California, where Governor Schwarzenegger established a nonpartisan commission to draw district lines, preventing that state’s Democratic legislature from gerrymandering there as well.

        I oppose gerrymanders, but if the Republicans are doing it, the Democrats should be able to as well. Since the Supreme Court gave their blessing to partisan gerrymandering, what’s needed is a lawsuit against CA and NY claiming that the nonpartisan districting discriminates in some way.

      • Azetheros  On November 16, 2022 at 8:06 am

        Re: Fetterman: I know someone who campaigned for him, HARD, in Philadelphia. From what I hear, Fetterman’s strategy was to focus on running up his margins in the rural counties where his image and story might be more relevant, and to let the party do its work for him in the cities. And, importantly, except for Conor Lamb and his people, all of the other primary candidates and all of the other wings of the party came together unreservedly to make that work. No one (else) decided to be petty in defeat.

        Now, I don’t want to suggest that ALL it took Fetterman to win was being a progressive (and I’ll at least embrace him as a progressive, fracking aside). It was a combination of factors that included his progressiveness, his image, and his strategy. But I think Fetterman’s campaign serves at least as a proof of concept for how a progressive candidate, with internal party support, can win statewide office. That you don’t *need* to run to the right to win like Fetterman did.

        And I REALLY hope his win causes the DSCC and the DCCC to reconsider how they recruit candidates. Fetterman is a candidate who could only emerge naturally through a primary process. He was NEVER going to be someone Chuck Schumer’s DSCC would have tapped, and I hope the lesson Schumer and other folks in leadership take is that they should stop trying to pick winners (colloquially speaking) centrally.

        Re: New York: One thing I’ve seen noted as further evidence that at least part of what happened was that Adams reinforced and legitimated the Republicans’ narrative about crime was a note on Democrats’ performance in Pennsylvania. Apparently Democrats outperformed Biden across the state, except for the corner of the state that is within the NYC media market. This remains a crude measurement, of course, but that plus Maloney’s absolute incompetence and selfishness, seems at least like a valid hypothesis to explore.

    • Guest  On November 16, 2022 at 3:28 pm

      Your two answers, Azetheros, are yes, absolutely, Dems leaning left helps, and two, Dem leadership needs to change in so far as they need to really hear the first answer.

      These are really tough questions for the Sift because Doug is a centrist down to his bones (and Doug, if you’ll allow, I’ll jokingly blame PTSD from McGovern in ’72, compounded by PTSD from Reaganism leading to the Clintonion “third-way is the only way” for you seeming to be stuck in the middle).

      In order to help Dems who still bristle at “progressive” perhaps we can frame it as delivering on behalf of the material needs and interests of ordinary people. Or at least, convincingly working in that direction, even if it falls short of what progressives would want. In fact, this is what Biden has done in the last year. Biden has been the most worker friendly pro-union president in decades, he on-shored jobs, threw a slice of bread to the base with very modest pot reform, and of course made big waves with student debt relief. It sounds unbelievable, but Biden in this past year has been the most progressive president that Gen X/Millenials/Gen Z has ever seen. Not in rhetoric (Obama keeps that crown) but in actual policy.

      If we see Dem leadership taking on massively popular progressive stances in their candidates, particularly economically populist ones, I’ll bet we see more Fettermans. Double-down on milquetoast centrist candidates, and we’ll see more of what New York got this round. It may be hard to remember with election deniers and Roe rightfully soaking up attention, but absent those concerns (again, New York may be a window into this, as there was no credible threat about abortion or elections being taken away there), we are living in Gilded-Era-level wealth inequality. People really are hurting out there. And they can see through corporate yes-men like Maloney.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: