Original story by Sam Luks and Brian Schaffner for The Washington Post.

The Republican Party often treats discussions of sexual harassment or discrimination as being of minor importance, while increasingly the Democratic Party treats women’s rights as central to its platform.

This shows how over the last several years, debates between Republicans and Democrats have increasingly involved extremely divergent attitudes toward women.

But behind that distinction, Democratic voters’ attitudes vary widely. In a general election, people generally vote based on party loyalty, mostly setting aside any other biases. That’s not a factor in a primary, which leaves room for voters to make choices based on stereotypical beliefs about people based on sex, race, religion, and so on.

Are attitudes that can be measured as “sexist” — meaning beliefs that women are less capable than men and overly sensitive to slights — influencing Democratic voters’ choice about whether to support a female candidate in the 2020 Democratic primaries? For instance, when Democrats worry about whether a candidate is “electable” — i.e., whether others would vote for him or her against Donald Trump — are they actually voicing their own discomfort about voting for women? Do these voters in reality see female candidates as inferior to male alternatives?

Sam Luks and Brian Schaffner for the Washington Post set out to answer these questions.

Twice, Luks and Schaffner surveyed 602 Americans online who said that they would be voting in the Democratic presidential primary or caucus in their state in 2020, via YouGov. Luks and Schaffner first interviewed a larger nationally representative sample of Americans in September and October 2018, having them answer a series of questions that are part of the hostile sexism battery — a scale developed by social psychologists Peter Glick and Susan Fiske to measure prejudice and hostility toward women. This scale asks people whether they agree or disagree with statements such as “women are too easily offended,” “most women interpret innocent remarks or acts as being sexist” and “most women fail to appreciate fully all that men do for them.”

When we combine six hostile sexism items into a single scale, we get a picture of how sexist Democratic primary voters are compared to other Americans. This is shown in the plot below. Perhaps unsurprisingly, most Democratic primary voters score lower on hostile sexism than other Americans. But among Democratic primary voters, there’s quite a wide range in sexist attitudes. In fact, more than one-fourth of Democratic primary voters score higher than the average American adult on the hostile sexism scale.

Luks and Schaffner then re-interviewed these Democratic primary voters in June 2019, asking them to indicate all the candidates they were considering voting for and to tell us which one was their top preference. Their answers were similar to what other polls found before the first debate: Joe Biden was the top choice for 33 percent of primary voters, followed by Bernie Sanders and Elizabeth Warren with 17 percent each, Kamala D. Harris with 8 percent, Pete Buttigieg with 7 percent and a host of other candidates at 3 percent or less.

So did their attitudes toward women influence their candidate preferences? To answer this, Luks and Schaffner divided the Democratic primary voters from our sample into four groups, based on where they fell on the sexism scale. For each group, Luks and Schaffner examined vote preferences — while controlling for a host of other demographic and political factors, including partisanship, ideology, gender, age, race, and religiosity. The figure below shows the results.

As you can see, levels of sexism appear to heavily influence Democratic primary voters’ candidate choices. Among the least sexist voters, Biden and Warren are neck-and-neck; among the most sexist Democratic primary voters, Biden is preferred by as much as a four-to-one margin. Warren’s support drops from nearly 30 percent among the least sexist voters to less than 10 percent among those who are most sexist. Harris’s support drops from around 15 percent among the least sexist voters to less than 5 percent among those who are most sexist. Thus, in both cases, the top female candidates lose about two-thirds of their support as they move from the Democratic primaries’ least to most sexist voters.

This effect goes beyond Warren and Harris. The figure below shows all the candidates in the field in our analysis — 17 men and six women (Tulsi Gabbard, Gillibrand, Harris, Klobuchar, Warren and Marianne Williamson). Because there are more men than women in the field, even the least sexist Democratic primary voters are considering more male candidates. But as sexism increases, Democratic primary voters generally consider fewer female candidates. Specifically, about 40 percent of the candidates considered by the least sexist Democratic primary voters are women, but among the most sexist Democratic primary voters that figure drops below 20 percent.

As progressive as many Democratic primary voters are on gender attitudes, there are still many people who will vote in Democratic primaries and caucuses next year who actually hold quite sexist attitudes. In fact, people with above average levels of sexism compared to the rest of the country could make up more than 25 percent of all Democratic primary voters. These individuals’ attitudes mean that the women running for the Democratic nomination will have a harder time winning than the men.

But what might this mean in the general election? It’s possible that some Democratic voters who are not currently considering any female candidates might choose not to vote for a female Democratic presidential nominee in the general election. However, sexist attitudes cost Republicans more votes than it gained them in the most recent midterm elections, as experiencing the first two years of Trump’s presidency pushed less sexist Americans toward the Democratic Party in 2018. So while sexism may be a hurdle for candidates like Harris, Warren or Gillibrand as they compete for the Democratic Party’s nomination, it could help them win a general election campaign against Trump.

Sam Luks (@scluks) is managing director of scientific research at YouGov.

Brian Schaffner (@b_schaffner) is Newhouse Professor of Civic Studies in Political Science and Tisch College at Tufts University.