The Funny Math Behind Centrist ‘Electability’ Arguments

The Revolving Door Project, a Prospect partner, scrutinizes the executive branch and presidential power. Follow them at therevolvingdoorproject.org.

In the Trump era, centrist politicos and pundits have repeatedly relied on two things to contain the surging left flank: gobs of corporate money, and “electability,” the argument that moderates are better suited than progressives to unify an anti-MAGA coalition. Progressive candidates overcame corporate money to win primaries in close elections, and in two weeks, the midterms will test the claim that only moderates can carry majority-making seats. 

Since progressive nominees have joined Democrats of every ideological stripe to lead in polls across the country, centrists have shifted their argument away from “progressives can’t be our candidates because they’ll lose important races” to “progressives win by less than they should.” This shift implicitly concedes that if there is a performance benefit to moderation, it is small enough to not matter, at least in this election. 

Michigan Senate candidate Abdul El-Sayed has been one of the major targets of this argument. In a recent Twitter post, Jack Miller, communications director at Chamber of Progress (the vehicle through which Big Tech tries to speak Democrat), wrote: “El-Sayed is underperforming in a D +8 environment. If he was a stronger candidate, we would have had more money for FL or MS, but instead we’re investing in MI.” 

Even accepting this frame, the chattering class seems to have far less to say about underperformance from establishment candidates like Chris Pappas and Cinde Warmington in New Hampshire or Kathy Hochul in New York. The think pieces about why these figures can’t be the future of Democratic politics are in conspicuously short supply. But asymmetry is a feature of the electability discourse, which increasingly rests on a particular method of evaluating electoral performance, a niche statistic called “Wins Above Replacement,” or WAR.

FIRST POPULARIZED BY THE POLITICS BLOG Split Ticket, WAR attempts to quantify the long-nebulous concept of “candidate quality.” It seeks to measure how much a candidate over- or underperforms the expected vote share of a generic, replacement-level candidate. Much of this (though not all) is calculated based on the presidential vote in the same district, which right from the outset is a questionable yardstick. 

The presidential race may arguably be the best baseline available for statistical comparisons, but it also isn’t a neutral comparison; presidential candidates don’t campaign or advertise uniformly across states, and they often have different regional and demographic bases. If Josh Shapiro were to be the 2028 nominee, anyone running in Pennsylvania would likely have a worse WAR score because of Shapiro’s strength in the state. Conversely, everyone who runs in places like Montana and Idaho, which the presidential campaigns largely ignored, will have an inflated WAR because of a lower baseline.

Split Ticket’s (and all other outlets’) WAR is markedly different from the stat that most sports statistical nerds are familiar with. In Major League Baseball analysis, WAR estimates how many games a team wins with a given player rather than with a replacement-level one (an average free agent or minor-leaguer) at the same position. Dodgers star Shohei Ohtani, for instance, currently has a WAR of 6.5, meaning the Dodgers win 6 or 7 more games a season with him over a replacement-level player. 

But a political WAR of 6.5 obviously does not mean that a candidate will win 6 or 7 more elections; in baseball, wins are a continuous variable, whereas in elections, a win is a binary outcome. 

A Split Ticket WAR of 6.5 actually means that a candidate’s vote share relative to the presidential vote in the same district is 6.5 points greater than what the group’s model predicts. Let’s say Abdul El-Sayed wins the Michigan Senate race with a final vote share of 52 percent. That’s about 4 points higher than Kamala Harris’s vote share in the state in 2024. But let’s say the model actually expected El-Sayed to get 55 percent of the vote, a predicted 7 points over Harris. That would make his WAR -3 (you can read the methodology here).

So WAR doesn’t say how much more or less likely a candidate is to win, only whether they beat the model’s expectation of their margin. Intuitively, a candidate who overperforms on that metric probably has somewhat higher odds. But the chance of winning is what actually matters. To borrow again from sports analysis, “a win’s a win.”

There has also been some criticism, chiefly from data journalists Mark Rieke and G. Elliott Morris and political scientists Jake Grumbach and Adam Bonica, that raises more direct methodological issues with Split Ticket’s metric. Bonica and Grumbach argue that the outlet massages data to favor moderates through proprietary “adjustments.” Morris and Rieke point out that the presidential vote share is a bad basis for candidate quality comparison because the nature of running for president is very different from running for Senate or House.

An alternative measure from Rieke’s The Data Diary and Morris’s Strength in Numbers that aims to address this exact issue is Wins Above Replacement in terms of Probability or WARP, which calculates how much more likely a particular candidate is to win a race than an average electoral performer. In essence, this approach can “separate margins that matter from margins that don’t.” An extra point of vote share is absolutely huge in an electoral nailbiter but immaterial in a rout. 

After the 2024 elections, a common centrist talking point was that progressives had “underperformed,” demonstrating that a more moderate party would have done better, perhaps even carried the elections overall. Sen. Elizabeth Warren, ever a centrist bugbear, was a favored example. She got 59.8 percent of the vote in her Senate re-election in Massachusetts. But Split Ticket assigned her a WAR of -10.8, saying that where Warren trailed Harris’s vote share by 2 points, the baseline should actually be running nearly 9 points ahead of the top of the ticket. By this measure, a Massachusetts senate candidate would need to win roughly 70 percent of the vote to not be an “underperformer.”

Strength In Numbers calculated Warren’s WARP at 0 percent—there was practically a 0 percent chance of Democrats losing that seat with her or with a replacement-level candidate. 

One problem with looking at WAR instead of WARP is that the Warren race was an un-campaign. She spent no money, made few public appearances, and in fact gave away much of her campaign war chest to candidates in close contests, while traveling the country to campaign for them. Her opponent got millions of dollars from the crypto industry, which may have uselessly moved the needle a bit. But this was in no way the kind of race we’re seeing Adam Hamilton or Annie Andrews run this year in deep-red states; it was never in doubt.

One key point that WARP captures and WAR misses: it is better for the party when candidates in safe seats do not run up the score. With the outcome essentially certain, it is smarter for a candidate in Warren’s shoes to not spend undue money on ads and canvassers, not compete for a limited pool of campaign talent, not stay almost entirely in-state rather than stumping for other candidates around the country, and not compete for advertising spots with Harris in the Boston media market, which also reaches the most population-dense area of the more-competitive state of New Hampshire around Manchester and Nashua. 

CENTRISTS LIKE MATT YGLESIAS AND ADAM JENTLESON of the Searchlight Institute have been laying the groundwork to argue that the performance of candidates like El-Sayed—at least since efforts to outright scuttle his campaign fizzled—and Troy Jackson in Maine is better judged by the final WAR scores than by who is sworn in to the 120th Congress. That is both a desperate scramble to keep the moderate “electability” talking point fresh well after its expiration date, and a terrible way to evaluate political success. 

Winning a race for a United States Senate seat is a monumental accomplishment, full stop. The power and privileges of a senator do not change depending on whether they win by 5 points or 0.5 points. 

What the “underperforming” tag misses is what El-Sayed did to reach this point. He defeated the full weight of a Democratic establishment that unified behind Haley Stevens, who racked up endorsements from retiring Sen. Gary Peters, Gov. Gretchen Whitmer, and Senate Minority Leader Chuck Schumer, among others. Stopping El-Sayed has also been a top priority for AIPAC all year long; they put $30 million into the effort, part of a total blitz against El-Sayed in the primary that reached nearly $70 million. 

While Stevens and most other Democrats rallied to his side after the primary, there are still centrist saboteurs trying to scuttle El-Sayed and elect Mike Rogers. The RNC has disseminated a “pro-democracy, anti-extremism” voter guide that includes an endorsement of Rogers by the Pennsylvania-based Beacon Coalition, which received a $200,000 grant from Third Way’s “Moderate Power Venture Fund” to oppose “the rise of extremist ideological pressure in the Democratic Party by building a pipeline of principled local leaders committed to pragmatism, pluralism, and responsible governance.” 

Third Way has since publicly disagreed with the Beacon Coalition’s actions. But that doesn’t change the fact that their centrist slush fund is helping to fund a group that seems not to be especially interested in actually fighting the Trump regime; it has also endorsed the Republican challenger to Rep. Summer Lee. 

If El-Sayed wins, it will be a major victory. An avowed progressive running on Medicare for All and weathering a fusillade of attacks on his religion to win statewide in a swing state flies in the face of the argument that progressives cannot compete in such environments. Still, regardless of the result, centrists will be waiting in the wings to label him an underperformer based on a model that factors in few of these headwinds. 

It is possible that progressives lose, making the centrist case stronger. But any rush to delegitimize a winning coalition for prevailing on its own terms is not neutral statistical analysis, but a rather obvious tactic to win a factional fight and check the party’s progressive flank. If and when you see such arguments, remember that old adage about lies, damned lies, and statistics. 

Just remember why politicians run for office in the first place (or at least why they should): to serve the public. As El-Sayed himself said, “You think my goal is to win—only? That’s not my goal. My goal is to govern well. We treat getting elected as if that’s some sort of achievement. That’s when the thing starts.”