There’s Probably a Small Gender Penalty. A Woman Could Still Have the Best Chance.
Matthew Yglesias recently pushed back against the trope that Democrats should nominate "a straight white Christian man" for the 2028 Presidential Election. He argued that women, Black, Hispanic, gay, and Jewish candidates can and do win tough races, academic research on gender and electability broadly supports this, and limiting the field to straight white male Democrats leaves you with a very short list.
Yglesias cites academic research to make this case. Multiple scholars have found that once you control for incumbency, party, and district partisanship, women win downballot races at statistically indistinguishable rates from men. Danny Hayes and Jennifer Lawless find that women are less likely to run, but conditional on running, they do just as well. Kathleen Dolan finds that gender stereotyping and sexism are real, but tend not to matter much to actual voting decisions. Deborah Jordan Brooks finds that gender stereotyping is more prevalent with political neophytes, but that on net women receive a moderate benefit from it.
Yglesias is right that the reflexive retreat to a straight white male candidate is strategically sloppy and probably wrong. But his implication that the downballot evidence settles the presidential question is not warranted.
There have been presidential elections every four years since 1788. No woman has ever won. That's decisive evidence that structural factors have kept women from becoming president.
The obvious rejoinder is that structural barriers have been collapsing — women now earn law degrees at rates comparable to men, have made substantial inroads in medicine and academia, and win Senate seats and governorships. All of this is true. None of it is dispositive. The question isn't whether women can succeed in elite professional life or even in most areas of electoral politics. The question is specifically whether a woman can win a presidential election. Those are not the same question.
The research Yglesias cites is almost entirely drawn from congressional and state races. That's where the data is, because there are hundreds of those elections every cycle. But presidential elections are a different animal. Voters think about the commander-in-chief differently than they think about a senator or a governor. The sample size for women running competitively for president is two. No serious statistician draws strong conclusions from that — though it's worth noting that both elections were close. If they had both been blowouts, the effect size might be large enough to overcome the small sample. They weren't.
Intuitions don't fill the gap. Smart, politically engaged people think they understand the electorate because they understand partisans, coalition dynamics, and the concerns of various interest groups. But the electorate isn't a single coherent thing. It's millions of subgroups making decisions through a mix of conscious reasoning and subconscious bias that no one has fully mapped. If one in twenty-five swing voters carries a residual, perhaps unconscious taste for male chief executives, that's a material electoral headwind. You wouldn't see it directly in polling. You might not see it in focus groups. You would only see it in outcomes.
The old patriarchal edifice has been substantially eroded. But "substantially eroded" is not the same as "gone," and we don't know how much of it persists specifically in presidential elections, specifically among the marginal voters who decide them.
Consider the following thought experiment. Title VII was passed in 1964. If it had annihilated the gender penalty and created a 50% chance of female victory in each subsequent election, there would be a 99.9985% chance that a woman would have won one of the 16 presidential elections since. If the penalty had disappeared as recently as 2006 — by which point women's professional and political gains were already substantial — there would be a 96.9% chance a woman had won one of the five elections since. If it had disappeared by 2014, the odds would still be 87.5% across the three elections since. None of those things happened. A gender penalty clearly persisted well into the recent past, and the burden of proof is on those who claim it has now vanished entirely.
One might object that even without a voting penalty, women have been less likely to run — and that suppressed candidacy rates, not voter bias, explain the historical record. This is worth taking seriously. Suppose the electoral penalty vanished in 1965 but women were only half as likely to emerge as major party nominees. Then the probability of a woman winning any given election drops to 25% — a 50% chance of running multiplied by a 50% chance of winning. Under those assumptions, the probability that at least one woman would have won across 16 elections is still 99%. For the five elections since 2006 it falls to 76.3%, and for the three since 2014 to 57.8%. The long historical record remains damning under almost any reasonable assumption about candidacy rates. The recent elections are less conclusive — but they are also the ones where women have actually run, which makes the candidacy-suppression objection weakest precisely where it would need to be strongest.
Anyone who tells you they know the answer — whether confidently predicting that gender is no longer an obstacle or confidently arguing it remains disqualifying — is reasoning from intuitions they haven't examined.
If forced to a number: a 1.5-point female penalty in presidential elections — measured in margin of victory — is a reasonable central guess. But it's worth being precise about what "zero" means here. The true effect being exactly zero is essentially impossible — in any real social phenomenon, exact nulls don't occur. If you flip a coin a million times, the probability it comes up heads exactly 500,000 times is less than 0.1%. Zero is simply the number that good progressives who support gender equality want the answer to be. Insisting on such a pristine outcome betrays either amateurism or motivated reasoning. The honest range runs somewhere between zero and three points.
We have run this experiment exactly twice under modern conditions, with sample sizes too small to be conclusive, on an electorate we understand imperfectly. That's not a reason for pessimism. It's a reason for epistemic humility — which points toward a practical conclusion.
Whatever penalty exists, it is small enough that it resists measurement. Preemptively disqualifying every female candidate on that basis isn't tactical sophistication — it's superstition dressed up as strategy. The voters who would quietly balk at a female president are unlikely to hide this from head-to-head polling; they'll simply back her opponent. That signal is already in the data, if it's there at all.
We have a recent cautionary tale — though not quite the one people think. Hillary Clinton entered 2016 with high negatives. Her defenders attributed this to misogyny, and misogyny may have been part of it. But that framing discouraged Democrats from taking the underlying numbers seriously as an electability signal. Had they looked at the publicly available polling data rather than explained it away, they might have made a different choice in the primary. The lesson isn't that Clinton lost because she was a woman. The lesson is that explaining away negative polling with unfalsifiable claims about voter bias is a way of not doing politics.
Democrats should run an open and vigorous primary, and voters who care about winning should pay close attention to head-to-head general election polling. The decisive question is not whether a gender penalty exists but which candidate, male or female, is the strongest. The polling can answer that. No one's intuition can.