Here’s a prediction that 100 percent, absolutely, positively will come true: I will get something wrong in 2023. Here at FiveThirtyEight, we make a lot of predictions every year; some of them work out, but we can’t get every single one right. 

We can, however, learn from our mistakes. That’s why I like to write about everything I got wrong in the previous 12 months.¹ I do this for two reasons: First, they’re often unintentionally hilarious (and when you’re a politics reporter, sometimes you need a laugh); second, identifying my blind spots has helped me become a better analyst.

And there’s no shortage of material for this year’s installment. Let’s start with a tweet I wrote on Nov. 6, 2020, shortly after it became clear that Joe Biden had won the presidential race: “Congratulations to Republicans on their victory in the 2022 midterms!” This was obviously meant to be snarky but also to communicate a political tenet: that the president’s party almost always has a bad midterm election. Of course, that tweet wasn’t from 2022, but I also made this argument in January of this year. And for several months thereafter, my analysis was colored by my expectation that 2022 would be a good election year for Republicans.

As everyone knows by now, the midterms were a disappointment for Republicans. They won the House — but only barely (they gained just nine seats on net). Meanwhile, Democrats gained a seat in the Senate.

140 million Americans will live in states controlled by Democrats | FiveThirtyEight

Clearly, I was overly confident in my early prediction. While it is true that the president’s party almost always has a poor midterm, there have been exceptions. And the 2022 midterms turned out to be one of these “asterisk elections,” thanks in no small part to the Supreme Court’s decision in Dobbs v. Jackson Women’s Health Organization to overturn the constitutional right to abortion. This year I should have been more prepared for the possibility that the ruling could throw a wrench into the election, especially after a draft of the decision was leaked in May. And even after the decision, it took me a while to become convinced that voter anger over Dobbs would prove durable enough to last until Election Day. It wasn’t until the fall that I revised my expectations from a “red wave” to a “red ripple.”

My biggest mistake here was not realizing just how common an “asterisk election” actually is. I often quoted one key stat: that the president’s party had gained House seats in only two of the previous 19 midterm elections. But there were four other midterms where the president’s party lost fewer than 10 House seats — so what happened in 2022 isn’t that rare. I also neglected to remember that the president’s party had lost Senate seats in only 13 of the last 19 midterms. In other words, midterms like 2022 happen about a third of the time — way too frequently to count them out.

I also spilled a lot of ink this year on congressional redistricting — some of which I regret. In March, I wrote an article that said the redistricting process was almost over, and I went on to analyze the congressional maps that had been passed at the time as if they were final. “The House Map’s Republican Bias Will Plummet In 2022,” I declared. I estimated then that the House’s new median seat would have a FiveThirtyEight partisan lean² from 1.0 to 1.9 percentage points more Republican-leaning than the nation as a whole.

That article was premature. It was another three months before the last state (Louisiana) finalized its map. In the meantime, New York’s congressional map was struck down in court, and Florida passed a map that exceeded partisan Republicans’ wildest dreams. When the dust settled, the national congressional map’s long-standing GOP bias had barely changed — the median seat was 2.5 points more Republican-leaning than the nation.

But then an interesting thing happened in the 2022 election: The congressional map benefited Democrats. Based on 2022 House results, the median seat³ actually voted 1.9 points more Democratic than the nation as a whole. You can add this to my list of misfires if you want, but I don’t think it disproves that the congressional map still has a Republican bias over the long term. This was a highly regionalized election, and turnout was higher in safely Republican areas than in safely Democratic areas, skewing the national popular vote. (Democrats also left more seats uncontested than Republicans did.)

What’s more, North Carolina and Ohio must draw new congressional maps for the 2024 elections. And Republicans won a majority of seats on both states’ supreme courts in the 2022 election. So the Republicans who control redistricting in those two states will likely be able to get away with strongly gerrymandered maps. As a result, I think the overall House map will return to helping Republicans again in 2024. (Probably.)

Over the course of the year, I also whiffed on several individual elections. For instance, I thought Rep. Madison Cawthorn would survive his primary challenge. But his many scandals caught up to him, and he lost 33 percent to 32 percent. I also called Rep. Jody Hice the favorite in the Republican primary for Georgia secretary of state. But Georgia Republicans surprised me by renominating incumbent Brad Raffensperger despite his defiance of former President Donald Trump. And when I first previewed Alaska’s special election for U.S. House in June, Mary Peltola was the ninth candidate I mentioned. But she pulled off the upset and won, becoming the first Democrat to win a statewide election in Alaska since 2008 and the first Alaska Native elected to Congress.

A few of my early general-election predictions aged like unrefrigerated mayonnaise too. In May, I wrote that Nevada Gov. Steve Sisolak had a distinct personal brand that could help win reelection. It turns out not so much: Sisolak lost to Republican Joe Lombardo. I also predicted that the governor’s race in Rhode Island could be sneakily competitive. Instead, Gov. Dan McKee ended up winning by almost 20 points.

Closer to the election, I also penned an article noting that Democrat Mandela Barnes was leading in the few polls there were of Wisconsin’s U.S. Senate race. Well, the day the article was published, Barnes lost his lead in our polling average and never regained it. Finally, after Hurricane Ian hit Florida, I was skeptical that Gov. Ron DeSantis’s response would help him politically. But in fact, DeSantis’s average polling lead went from 6 points on the day Ian made landfall to 12 points on Election Day, and he wound up winning by 19 points.

The prediction that hurt the most — at least, when it came to my post-election social life — wasn’t about who would win an election; it was about when someone would win it. In our interactive guide to election-results timing, published on Nov. 5, I wrote that we would likely know the winner of the election sooner than we did in 2020. How naive I was: It again took until the Saturday after Election Day for the networks to project a winner in the Senate — and eight days after the election for them to project the House. Oh, sweet Nov. 5 Nathaniel.

We answer your lingering questions about 2022 | FiveThirtyEight Politics Podcast