Heading into the election, several national indicators pointed to the Republican Party likely having a fairly good night. Biden’s approval sat at 44%, the generic ballot sat at R+1.1, and inflation was still around 8%, with gas prices more expensive than they were during the red wave-esque environment of November 2021.
One key problem for the GOP, however, was they faced heavy deficits in statewide polling conducted by high-quality pollsters, with a host of controversial nominees dragging down the party’s image in crucial battlegrounds, leading to a risk of Republican underperformances in key swing states like Georgia, Arizona, and Pennsylvania, which would collectively decide Senate control. The main question in our minds was to what degree this candidate effect would materialize, given the declining importance of candidate quality over the last few years. While we believed Democrats would eke out victories in Arizona and Nevada’s Senate races, and that Raphael Warnock would get to a runoff because of a Herschel Walker underperformance (especially when compared to fellow Republican Brian Kemp), we did not think it would be enough to let John Fetterman win against Dr. Oz in Pennsylvania.
While we called all of the other 34 Senate races correctly, we were clearly wrong on Pennsylvania’s Senate election and the degree to which candidate effects mattered there, and it was the lone missed state in our pre-election predictions. More broadly, while we expected that a Democratic overperformance was well within the bands of expectation, as shown by our generic ballot tracker and a host of Democratic results in the post-Dobbs special elections, we definitely did not expect that it would be on the backs of a rise of ticket splitting and heavily localized state and district effects, as ended up being the case.
So, what happened? How did Democrats manage to have the best midterm result for an incumbent party in twenty years? The answer, simply put, lies in persuasion, and arguably nothing reflects this better than the systematic Republican underperformance in the key Senate battlegrounds.
Republican Senate nominees in this cycle gained infamy for adopting a host of controversial positions, including advocating for a federal ban on abortions and questioning the validity of the 2020 elections. This seems to have cost them quite significantly in a host of states, as shown by the table below, and it appears to have tanked their performance in many swing states. In every battleground state, they underperformed significantly when comparing the actual result to the state’s environment-adjusted state lean (which is how a state should vote in an R+1 year, as 2022 looks likely to end up as, when using 2020 results as a baseline). Moreover, these underperformances were generally significantly more severe than explainable by incumbency-related effects.
Simply put, the above results are the strongest possible rebuke to the case that candidates do not matter. Republicans did not overperform the environment-adjusted partisan lean in any of the 9 battlegrounds rated as competitive by Split Ticket — perhaps not coincidentally, they ran election deniers in 8 of these 9 states. Democratic overperformances were not uniform — in states like Connecticut, California, and Florida, they appear set to underperform quite significantly. But in swing Senate races with heavy amounts of spending and campaigning, it is clear that the Democratic ticket significantly overperformed in virtually every case.
Given that Democratic turnout was actually quite weak in most states and lagged significantly behind Republican turnout, it appears as if Democrats flipped many Republican-leaning voters who were unenthused by the GOP nominees. This is not a novel insight — Nate Cohn of the New York Times found that the “MAGA penalty” for extremist candidates was 6.6 points on margin. But our table simply adds to the growing well of evidence that Republicans are beginning to pay a very significant candidate quality penalty which is costing them in many key races.
In a chamber like the US Senate in a year like 2022, there is no reason that a slate of generic Republicans should be losing seats. The problem is that the GOP do not run these candidates and frequently appear to nominate problematic and often-unelectable ones instead. Voters appear to have rejected the gambit this time, and while the Senate map still leans Republican in 2024, with Montana, West Virginia, and Ohio up for grabs, the GOP will want to fix this problem if they can.
The importance of candidate quality was just as visible in the gubernatorial elections, where Split Ticket missed only two out of 36 races (Wisconsin and Arizona). The strongest-performing Democratic gubernatorial candidates were moderate incumbents Laura Kelly, Jared Polis, Gretchen Whitmer, and Tony Evers. Another notable overperformer was Pennsylvania’s Josh Shapiro, who beat Joe Biden’s 2020 margins by 14 points in an open race. Part of this is attributable to his own candidate-specific strength, as Shapiro was consistently the strongest statewide Democrat in both his 2016 and 2020 runs as Attorney General, but part of this was also undoubtedly due to far-right Republican opponent Doug Mastriano’s hopeless campaign that aired more conspiracy theories than television advertisements.
Meanwhile in Arizona, Republican Kari Lake’s far-right embrace of bizarre and baseless election fraud claims provided Secretary of State Katie Hobbs (D) with an opening to win, which she successfully exploited. Nevada’s Steve Sisolak was the only incumbent governor to lose this cycle, as he was outmatched by one of the GOP’s best recruits this cycle in relatively uncontroversial Clark County Sheriff Joe Lombardo, who openly rejected Donald Trump’s baseless claims of election fraud. Note that the below table, like its predecessor, does not account for incumbency or spending (controlling for these would likely make candidates like Lombardo look a bit better and Newsom a bit worse).
Looking at the Republican overperformances, these can be generally attributed to two things. The first is popular incumbent Republican governors, such as Ron DeSantis (FL), Mike DeWine (OH), and Chris Sununu (NH). For the first two, the longstanding political trends are very much in Republicans’ favor, and in the Granite State, the Sununu name is political gold. The second is the lackluster Democratic campaign operations from candidates in deep blue states, such as Kathy Hochul (NY), Tina Kotek (OR), and Gavin Newsom (CA).
In New York, Kathy Hochul’s weak campaigning combined with the high salience of crime as an issue provided Republican Rep. Lee Zeldin with an opening to consolidate moderates and less-partisan Democrats to come within single digits of winning. In Oregon, outgoing governor Kate Brown’s unpopularity likely rubbed off on successor Tina Kotek, formerly the Democratic Speaker of the House, to some degree. In California, the likeliest explanation for Newsom’s underperformance would be his lack of a serious campaign, secure in the knowledge that he would win comfortably regardless.
Democrats also picked up governorships in deep-blue Massachusetts and Maryland due to the lack of credible moderate GOP candidates, thus breaking the moderate Republican streak that has been persistent in those state’s gubernatorial races for a while. The end result saw Democrats gaining three seats (Maryland, Massachusetts, and Arizona) and losing just one (Nevada), which is another somewhat unusual result in a midterm year under a Democratic administration.
Between a lackadaisical showing in the House, a disappointing year in gubernatorial races, and a horror showing in the Senate, Republicans significantly underperformed expectations this year. Voters cannot be taken for granted, and yet by nominating extremists in key races that consistently parrot nonsensical conspiracy theories, the GOP appears to have done exactly that. No political party wins forever, and there will almost certainly be a day again where the Republicans win, and win big. But to do that, they likely need to solve the issue of candidate quality that has once again dogged them in an election that they should have performed a lot better in. Given that this problem has now cost the GOP winnable Senate races in 2010, 2012, 2014, 2018, 2020, and 2022, observers may be excused for wondering if this tendency is a feature rather than a bug.
I’m a software engineer and a computer scientist (UC Berkeley class of 2019 BA, class of 2020 MS) who has an interest in machine learning, politics, and electoral data. I’m a partner at Split Ticket, handle our Senate races, and make many kinds of electoral models.