It’s no secret that the Senate is an extraordinarily difficult and uphill challenge for Democrats — in order to win the chamber in 2026, the party will likely need to flip at least two of Texas, Ohio, Iowa and Florida. Meanwhile, their hopes of competing in 60 seats, like they once regularly did and like Republicans still regularly do, are all but extinguished under current coalitions.
Enough ink has been spilled on the structural and coalitional forces (and choices) that led both parties to where they are today. But today, it’s worth looking at one somewhat-underdiscussed cause: the average voter simply doesn’t split their ticket nearly as much any longer, making the overperformances of the past a distant dream.
With the release of our 2016 Wins Above Replacement (WAR) models for the House and the Senate, we can now actually calculate the average impact of candidate quality per cycle. And the trend is extremely clear: since Trump’s victory, polarization has accelerated greatly, leading to a plunge in the average impact of candidate quality.
In 2016, candidate quality was worth almost 6% per race, in either direction — meaning that on average, a tied open seat would go to D+6 or R+6, based on candidate quality. By 2024, this had been sliced in half to just 3%, as local party and race differences began to fade in favor of nationalized politics. Challengers and incumbents alike now struggle to set themselves apart from the national party, and voters increasingly vote accordingly.
Polarization has also caused the band of plausible outcomes to shrink for any given race — 12 of the 14 elections with candidate effects of 20+ points happened in 2016 and 2018, and the remaining two happened in 2020 and 2022. By 2024, no election saw candidate quality yield a 20+ point electoral effect.
Importantly, this trend is even starker for 15+ and 10+ point overperformers: 59 races had a candidate effect of more than 10 points in 2016, but by 2024, this number had plummeted to just 13.
More than anything, this is the real reason Democrats have been shut out from nearly half of the nation’s seats: the decline in ticket splitting means they can no longer overcome structural coalitional differences as easily as they once did. In 2012, they were winning Senate races in states like North Dakota and West Virginia, even as Barack Obama lost them by more than 20 points. But virtually no modern candidate achieves that type of overperformance now.
To consistently compete on a broad map, Democrats will need +15 WAR candidates to flip red states like Kansas and Indiana — and it is now exponentially more difficult (though still not impossible) to find them. And while many may complain about the rural bias of the Senate making life tough for Democrats, it’s really the evaporation of ticket splitting that makes a broad Senate majority very, very tough for them, until perhaps the next structural change in American politics.
Discussing our 2016 WAR models
If you missed the brief mention in the section above, we’re excited to release our 2016 Wins Above Replacement (WAR) models for the House and Senate. If you missed the release of our 2024 model, here’s a quick look at how everything works: our models control for partisanship, incumbency, demographics, and a baseline spending threshold (which serves as a proxy for “competition”) to project a fundamental outcome for each race.
The fundamentals-based outcome serves as the estimate for how a race “should have” gone with a generic Democrat against a generic Republican, given the conditions of the race. Comparing this to the actual result allows us to quantify the impact of candidate quality across each cycle. For instance, if the fundamentals-based outcome suggests that a race should have been R+3, but the Democrat won by 1 point, then the WAR for that race would be D+4.
We found this to be the most fascinating set of explanatory WAR models we’ve ever created, and the reason is simple: ticket-splitting. It created novel challenges for us to model, especially as trends differed across demographics and geographies (for instance, congressional Republicans comfortably outran Trump with college-educated voters, but underperformed him with non-college whites). And it led to some of the wackiest findings we’ve ever seen, like Tom Price having a breakeven WAR despite winning a Trump +1.5 seat by 23, mostly because Democrats ran someone who arguably didn’t even exist.
You can find our full, downloadable WAR database here, aggregated across all years from 2016 to 2024. To our knowledge, this is the only database of its kind that exists, and we hope to keep maintaining and refreshing it.
In addition to the overall decline in ticket splitting, big changes in political geography have taken place during the Trump era. Looking at races from the 2016 cycle where candidate quality made the difference foretells some of these changes.
In the House, Democrats picked up two seats in Florida, defeating suburban Republicans who should have held on due to the strength of downballot lag. Similar losses occurred in New Hampshire (where candidate quality netted Democrats a Senate seat), Nevada, and New Jersey. Democrats would go on to win two dozen highly educated, high-income seats like these in the 2018 blue wave. Many of these districts, and the suburbanites residing in them, would not return to the Republican fold afterwards.
2016 also foreshadowed the solidification of Trump’s gains in rural seats. In Minnesota, Democrats retained just enough crossover support to hold on in three rural seats — two of which would flip in 2018 (despite the favorable national environment) following retirements.
Clearly, Democrats have consistently had the advantage when it comes to races where candidate quality ended up making the difference. This has been particularly true in the Senate, where Republican candidate quality woes have cost them a comfortable majority time and again (in 2020 and 2022, it actually cost them the majority altogether).
Our WAR models do collectively show that candidate quality matters and will continue to matter — even if it doesn’t hold as much sway as it did a decade ago. Just because crossover voting has declined does not mean that it is on the way out. There simply aren’t enough highly informed, straight-ticket voters for that to happen. For now, candidate quality (after controlling for other important factors) will be enough to make the difference at the margins — especially in swing districts and competitive states where majorities are made.
Editor note: following the 2016 WAR model release, we updated our 2018, 2020, and 2022 models to be consistent with this methodology in July 2025.
I’m a computer scientist who has an interest in machine learning, politics, and electoral data. I’m a cofounder and partner at Split Ticket and make many kinds of election models. I graduated from UC Berkeley and work as a software & AI engineer. You can contact me at lakshya@splitticket.org
My name is Harrison Lavelle and I am a co-founder and partner at Split Ticket. I write about a variety of electoral topics and handle our Datawrapper visuals.
Contact me at @HWLavelleMaps or harrison@splitticket.org
I’m a political analyst here at Split Ticket, where I handle the coverage of our Senate races. I graduated from Yale in 2021 with a degree in Statistics and Data Science. I’m interested in finance, education, and electoral data – and make plenty of models and maps in my free time.