Looking Back at the 2022 Projections



— After overestimating Republican performance in 2022, we wanted to give a short explanation to readers about our thinking in the run-up to the election.

— In the end, and with a lot of contradictory information, we thought the indicators pointed more toward the Republicans than the Democrats.

— In a political world where “lol, nothing matters” seemed to be a safe assumption in recent years, it appears that a lot of things did matter — things that should matter.

How we assessed 2022 the way we did

The day before the 2022 election, HuffPost’s Jonathan Cohn tweeted the following: “Feel like I could selectively pull anecdotes and data to make a convincing case for a big election surprise in the D direction — and then do the very same thing for a big election surprise in the R direction.”

One of us replied to the tweet, saying the following: “It’s true. I think in the end, a good R election is easier to explain/anticipate (fundamentals) than a good D one.”

In a nutshell, this is the best explanation we can offer as to why we thought Republicans would do better in this election than they did. With a lot of conflicting indicators, we went in the direction that history suggested was most prudent: namely, that President Biden’s approval rating and the usual midterm drag on the presidential party would push some — not all — of the key races toward Republicans. This overall assessment, which ultimately was a misread, ended up coloring our thinking as we made our final picks.

At various points of the cycle, we pondered whether the election would be more of a referendum than a choice. In the end, we edged more toward “referendum.” The reality was more “choice,” as voters rejected many weak and extreme Republican candidates across the country.

It does appear that, broadly speaking, the 2022 electorate was more Republican-leaning than 2020. Republicans are on track to win the House, albeit by a margin that could very well end up being smaller than the tiny 222-213 edge that Democrats won a couple of years ago. But Republicans just did not do as well as we thought was likely: In the end, they’re only going to end up gaining less than half of the 24 net seat gain we projected. In the 50-50 Senate, we gave a small edge to the Republicans; it went the other way, although we only missed Pennsylvania (and, arguably, Georgia, although because that race went to a runoff, everybody gets something of a do-over). In the gubernatorial races, we had 5 Toss-ups for months leading up to the election. In the end, we thought Republicans would get 4 of the 5; they ended up getting only 1.

All of these assessments really sprang from the same premise, which was that the political environment would be just a little bit better for Republicans than it ended up being — and that the political environment would manifest itself more evenly across the country than it did.

Part of what we do at the Crystal Ball is pick all of the Toss-ups at the end of the election. We do this because, ultimately, analysts who are focused on trying to game out election outcomes should make picks, we think. And because it gives readers a chance to judge how we did, without obfuscation. Some years are better than others — this one falls into “others.” We also aim to provide a lot more information about politics than just handicapping — including in-depth analysis of results, historical perspective, voter trends, redistricting, and much else — which we think, and hope, makes us worth reading beyond just analyzing the horserace.

In the days leading up to an election, we always canvass a number of sources on both sides. We conduct these conversations off the record — we are not looking for quotable information, we’re merely looking for data that can help inform the picks we make. The message we received over and over again, in the Toss-up House, Senate, and gubernatorial races, was that many races were extremely tight in final polling conducted on both sides. There were, to put it in the words of one Democratic source, a lot of “jump balls.” The worry for Democrats, as expressed by this source and others, was that the jump balls would end up disproportionately falling to the Republicans. This ended up being reflected in our thinking, even though we don’t think we forecast some sort of unequivocal “red wave” — we described our final projection as representing a “good but not necessarily great night for Republicans.”

Similarly, from what we were able to gather, while House Republicans had many promising targets, they had not actually put many of them away, which we noted in the leadup to the election. The results bore this out. Republicans picked off some incumbents in Trump-won seats in states like Arizona and Iowa, but, on the whole, Democrats in marginal Joe Biden-won districts fared well.

Collectively, we would describe the pre-election polling, both in public and in private, as inconclusive. That said, some major public pollsters who had some struggles in either 2016 or 2020 or both performed exceedingly well in 2022. We tip our caps to them. But what we struggled with, in advance, was believing these polls given their overstatement of Democrats in the past. That’s not intended to be some slight on these pollsters, and it’s a criticism one could just as easily level against us: We have sometimes overstated Democrats in the past in our own assessments, too, most notably in the 2016 presidential and 2020 House races. Our only point is that there were reasons to not necessarily take the good polls for Democrats at face value, just like there were reasons to be suspicious of the GOP-friendly pollsters who produced rosy numbers for Republicans in many races down the stretch. The end result, polling-wise, was something of a muddle, at least as we evaluated it. The final House generic ballot polls did indicate, generally, a small Republican edge — it appears that will end up being accurate.

Aside from polling, there were a few other indicators that we probably should have weighted more heavily when putting together our forecasts. After the Supreme Court’s Dobbs decision, Democrats overperformed in a string of special elections: starting in late June, Republicans won double-digit Trump districts in Minnesota and Nebraska by smaller margins; then, in August, Democrats flipped the Alaska At-Large seat and fared well in a pair of Upstate New York districts. Because special elections often happen in isolation, and usually see lower turnout, it is hard to project their results forward. But as one of us quipped, in the larger scheme of things, “special elections weren’t so special” this year. Immediately upon his election, we made now-Rep. Pat Ryan (D, NY-18) a favorite for reelection, and the Crystal Ball was the first major handicapper to move Alaska into the Leans Democratic category for now-Rep. Mary Peltola (D) — the former was reelected and the latter is in a strong position. But the special elections ended up being an important clue, particularly in regards to Democratic mobilization in the wake of Dobbs. Although it is interesting that, following the good results for Democrats in the specials there, New York ended up being a real trouble spot for Democrats anyway.

Last week’s results also seem like a vindication of the predictive value of Washington state’s late-summer primary. Usually held in August, it historically has had some value as a bellwether. In the Senate primary, Democratic candidates combined for 57% of the two-party vote — with much of the vote counted, that is exactly where Sen. Patty Murray (D-WA) is sitting now. In the primary, Democrats narrowly outvoted Republicans in the light blue 8th District, where Rep. Kim Schrier (D) won a third term last week. It seemed possible to us that the national environment may have deteriorated for Democrats in the months between August and November, but Washington state’s results proved prescient. The Washington state results were one of the few good indicators of 2020’s eventual House results, too.

In Pennsylvania’s Senate race, where we had Democrats favored all fall until shifting to Republicans at the last minute, we heard about very close internal polling on both sides. While we were not privy to this prior to the election, HuffPost’s Daniel Marans reported earlier this week that Sen.-elect John Fetterman (D), in his own internal polling, eventually fell behind Mehmet Oz (R). Ultimately, we didn’t think Gov.-elect Josh Shapiro (D) would produce the requisite coattails for Fetterman — based on our best info, we thought Shapiro would win by something like 10 points or a little less, which we thought gave Oz a better chance to ultimately get over the finish line. As it was, Shapiro is currently ahead by about 14.5 points, and Fetterman is up by about 4.5.

Our other Senate picks fared better. We had Wisconsin and North Carolina at Leans Republican all cycle — Sen. Ron Johnson (R-WI) won by about a point, while Sen.-elect Ted Budd (R-NC) held an open seat by 4 points. We also had Ohio rated as Leans Republican, where Sen.-elect J. D. Vance (R) underperformed Trump’s margins but still won 53%-47%. Despite some late drama, we stuck with Democrats in Arizona and New Hampshire, wisely (we never had New Hampshire as a Toss-up, and we moved Arizona to Leans Democratic in late August and kept it there). As it turned out, Sen. Mark Kelly’s (D-AZ) current 5-point margin would be the widest for a Democratic Senate candidate in Arizona since 1988, and Sen. Maggie Hassan (D-NH) outperformed Biden’s showing in the Granite State. We ended up getting Nevada right, too. We also never had states like Colorado, Florida, and Washington — all races that appeared as though they might be tightening at various points of the year — rated in a category more competitive than “Likely.” Our projections, including those we made in some of the closest and most pivotal races, got a fair amount right about the Senate.

There’s one other point we want to make, which we ultimately think speaks well of the American voter.

Throughout the cycle, it was an open question as to whether several weak and extreme Republican candidates would ultimately be punished by voters. Many parroted Donald Trump’s lies about the 2020 election, including several secretary of state candidates in key states. Some of these same candidates, or others, took positions on the highly salient topic of abortion that were far out of the public mainstream; others made flippant comments in the wake of, for instance, the violent attack on Paul Pelosi, husband of House Speaker Nancy Pelosi (D). A number of those candidates ended up losing.

As we wrote last week in the immediate aftermath of the election, “The GOP’s lack of discipline in candidate selection cost them a number of races.” The only way a party’s electorate and leaders will make better decisions is if they know that future mistakes will cost them future power. In a political world where “lol, nothing matters” seemed to be a safe assumption in recent years, it appears that a lot of things did matter — things that should matter. Key voting blocs in different states cast different verdicts and made what we think were fairly sophisticated judgments, as opposed to just issuing the kind of black and white verdict on the party in power that we sometimes get in midterms.

In the end, the electorate did do what it often does in a midterm — put a check on the White House. Narrowly, Republicans do look like they will elbow their way into formal power in Washington by flipping the House. But in denying Republicans the sweeping victory that many of them anticipated, voters rendered a mixed verdict. It’s one we will try to better understand and better explain as vote counts are finalized and as the 2024 presidential cycle begins — which effectively kicked off last night with former President Trump’s long-expected bid to return to office.