Our 2022 Senate Wins Above Replacement Model

During the course of our 2022 postmortems, we at Split Ticket have already quantified the importance of candidate quality in the House of Representatives elections. Today, we extend our model to the 2022 US Senate elections, where the disparities in candidate strength were even more striking.

The story of how the Democrats kept the Senate in 2022 can be told in one simple sentence: Democrats had really strong candidates, and Republicans nominated some incredibly weak challengers. We have made several wins-above-replacement models, and in many of them, the Republican Achilles heel of candidate quality has reared its head to varying degrees. but the 2022 GOP field of battleground challengers might be the weakest candidate field that we have ever seen a major political party put up in the modern era of American politics.

Our Wins Above Replacement model for the Senate shares many of the same characteristics as our 2022 House WAR model (and, in fact, is trained on the same dataset). It controls for a state’s demographics, its partisanship in 2016 and 2020, incumbency, and the financial disparities between candidates. Using this, we then evaluate the predicted margin and compare it to the real result. The results are displayed in the map below, and an interactive table of the full results can be found here.

There were 31 Senate races that we evaluated. In those, 19 races saw a Democratic (or Democratic-aligned) candidate overperformance, and 11 races saw a Republican overperformance, with one net-neutral result in New Hampshire. In an even more striking finding, out of the ten 2020 battleground states, 8 saw a Democratic overperformance, and the only one that saw a Republican overperformance was Florida, where Marco Rubio did 3.4 points better than expected against congresswoman Val Demings. In fact, our model suggests that under a national set of generic matchups, Republicans should have won 53 seats, but they instead ended up with 49 under the real ones — candidate quality was the differentiating factor in Nevada, Arizona, Georgia, and Pennsylvania and cost them the majority.

This probably does not come as a surprise to those who followed the cycle. Republicans nominated some very weak candidates who made some extremely controversial remarks and baselessly denied the results of the 2020 election. While this aligned themselves well with the GOP base, it put themselves squarely against the median voter’s preferences, especially on matters like abortion and election administration. This came back to bite them against an exceptionally strong slate of Democratic candidates, many of whom were incumbents. The inevitable sense of victory that pervaded the Republican camp proved to be nothing more than a mirage, and the lack of message discipline exhibited by the GOP came back to haunt them in a cycle transformed by both the Dobbs v. Jackson ruling and the specter of Donald Trump and the January 6th insurrection.

But if one can say that Republicans lost the Senate because of this this, then it is also true that Democrats won the chamber by doing the exact opposite. Democratic candidates exhibited relentless message discipline from the start, putting the party into the lead in states like Pennsylvania and Arizona while remaining competitive in seats like Ohio, and forcing the GOP to play defense in red states despite having a favorable map. Considering that every major battleground seat except Colorado and New Hampshire were more Republican than the nation overall in 2020, this was a necessary precondition to Democrats remaining competitive in the Senate.

The performance of candidates in battleground states (defined as states that ended up within 10 points in 2022 or states that voted within 10 points of the nation in 2020)

In Arizona, Blake Masters took a number of unpopular stances on birth controlgay marriageabortionsocial security, and the validity of the 2020 election. Unsurprisingly, none of these stances played well in a state won by Joe Biden, especially against an incumbent like Mark Kelly, who was the strongest victorious Democratic challenger in 2020. A generic Republican probably would have won this race against a generic Democrat, but the GOP needed a John McCain-esque candidate to unseat someone as strong as Kelly, and they paid the price for nominating the polar opposite. Kelly capitalized early on his opponent’s mistakes, used the airwaves effectively with the swaths of money he had raised, and sank Masters’ favorables from the start. Masters ended up losing by 5 points and underperformed by 6.6 against the incumbent Democrat.

In Georgia, the only way for the GOP to lose a heavily Republican 2022 electorate was to nominate an unelectable candidate. But this is exactly what ended up happening; if incumbent Democratic senator Raphael Warnock was the picture of electability, with a relentless focus on bipartisan cooperation and astounding message discipline, then Republican recruit Herschel Walker was the polar opposite. The ex-football star’s numerous abortion scandals and disturbing backstory, including domestic abuse allegations from his ex-wife and a bizarre flap over the number of undisclosed children he had, eventually torpedoed his campaign. Walker lost by 2.8, posting the biggest Republican statewide loss in nearly two decades, and underperformed by 6.2, blowing a race that many would have initially considered to be the political equivalent of a layup.

Walker and Masters weren’t the worst Republican Senate challengers, however. On paper, Mehmet Oz looked like a stellar recruit for the GOP in Pennsylvania, and the GOP hoped that the charismatic teledoctor could leverage his name ID and celebrity status to help them hold on to Pat Toomey’s seat against Lieutenant Governor John Fetterman. In practice, Oz’s campaign was hamstrung from the get-go by an exceptionally bloody GOP primary, a brutally unpopular Republican gubernatorial nominee, and a series of needless missteps, which kept his unfavorables extremely high throughout the cycle. Meanwhile, Fetterman hit Oz repeatedly on the carpetbagging angle and painted him as an out-of-touch elitist, all while simultaneously flooding the airwaves with lots of positive messaging. The end result was a borderline blowout — Oz lost by 5 points to Fetterman, underperforming by 8.9 points in the process and posting the worst open-seat battleground underperformance in possibly a decade.

In the aftermath of the elections, many Republicans pointed to the closeness of Nevada’s senate race and claimed Adam Laxalt was one to follow for the future. We are less sure about this assertion — Laxalt lost the same electorate that ousted an incumbent Democratic governor, and our modeled estimates actually suggest that Laxalt underperformed by 3.2 points. While it is true that Laxalt was not as bad of a candidate as Walker, Oz, or Masters, his performance was still extremely underwhelming against another strong Democratic incumbent in Catherine Cortez Masto. The Democratic portrayal of Laxalt as an election-denying Washington elitist kept the former Attorney General’s unfavorables high throughout the cycle, helping the incumbent Senator hold on in a Republican-leaning electorate.

Perhaps the most surprising finding for us came in New Hampshire, where our model estimated that Don Bolduc was really just a replacement level candidate despite him losing by 9 points in a Biden +7 state. This is likely because of two reasons: firstly, the model accounts for fundraising to some extent, and this is an area in which Bolduc was anemic — observers may credibly point out that this is also an area in which a candidate bears some responsibility, and we would not push back too hard on the assertion. Secondly, New Hampshire is one of the states that was projected to be extremely demographically favorable for Democrats due to its exceptionally white and secular electorate — indeed, the entire New Hampshire GOP slate collapsed federally, with Annie Kuster and Chris Pappas soundly defeating Bob Burns and Karoline Leavitt. Given this context, the model simply believes that other Republican candidates with Bolduc’s resources wouldn’t have done any better or worse than he did against an incumbent like Hassan, and we find this to be a reasonable conclusion.

Democrats appear to have had a candidate quality edge over the GOP in other battlegrounds too. In North Carolina, arch-conservative congressman Ted Budd defeated former State Supreme Court justice Cheri Beasley by 3, but our model suggests that he should have won by a few more points than he really did. In Wisconsin, meanwhile, controversial Republican incumbent Ron Johnson eked out a 1-point victory against Lieutenant Governor Mandela Barnes in a race that should have been a more comfortable 5-point Republican victory. And in Ohio, venture capitalist JD Vance beat Tim Ryan by just 6 points in a Trump +8 state. This was a poor showing that almost certainly had significant downballot effects, as Greg Landsman and Emilia Sykes won closely-contested battleground House seats. Even Mike Lee only won by 10 (and underperformed by 17 points) against Democratic-aligned independent Evan McMullin in Utah, who painted the GOP incumbent as an extremist and highlighted his pro-Trump alignment.

Republicans may consider themselves unlucky to not emerge with the majority in a midterm that saw the Democratic President’s approval at 42% and inflation at 8%. But given what our modeling implies about some of the candidates that they put up, perhaps they were also lucky to still end up with 49 seats. Fundamentally, candidate quality still matters — and while the Republicans had everything else on their side, they did not have this.


We hope you enjoyed this article. If you did, please consider subscribing to get our articles in your inbox — it’s free!

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.

I make election maps! If you’re reading a Split Ticket article, then odds are you’ve seen one of them. I’m an engineering student at UCLA and electoral politics are a great way for me to exercise creativity away from schoolwork. I also run and love the outdoors!

You can contact me @politicsmaps on Twitter.

Leave a Reply

%d bloggers like this: