:format(webp)/cdn.vox-cdn.com/uploads/chorus_image/image/49090263/GettyImages-514707186.0.jpg)
Presidential candidates often tell stories on the campaign trail about their families or people they've met. But strangely enough, almost all of those stories are about men.
Textio, which analyzes gender bias in job postings, decided to run the presidential candidates' stump speeches through its algorithm. The team examined the structure and content of 126,362 words of campaign speeches by Bernie Sanders, Hillary Clinton, Donald Trump, Marco Rubio, and Ted Cruz.
And they found that every candidate, even Clinton, tells a lot more anecdotes about men than women on the campaign trail.
Clinton and Trump were the most gender-balanced in their stories. Both told about three times as many stories about men as women — but more than 80 percent of Trump's references to women were about his own wife or daughter. Sanders referred to men about eight times more than women, Cruz eight times more, and Rubio a whopping 18 times more.
Textio also looked at the balance of "gendered" phrases the candidates used. Hillary Clinton's "average gender tone" skewed slightly on the feminine side, and all of the other candidates skewed masculine.
The designations aren't based on whether a candidate is male or female, but on each individual's choice of language. For instance, analysts compared the candidates with other high-powered public figures and found Martha Stewart spoke in a gender-neutral way, while Mark Zuckerberg's speech skewed ever so slightly feminine.
Clinton's speech is dominated by phrases like "balance work and family" and "communities are strong," while Cruz is more likely to say things like "absolutely destroy" and "defend at all costs." Those are pretty stereotypical-sounding examples, but not all of them are that easily guessed — for instance, "striving" is coded as feminine but "determined" is masculine.
Textio's model analyzes which words or phrases are more likely to attract male versus female job candidates. It's designed to help companies avoid using male-biased language in their job postings — and job postings are still what it does best, CEO Kieran Snyder told Vox. But its data set has also expanded to include language from emails, and users often plug in other kinds of text just to see how it measures.
In other words, feel free to take Textio's designations of what counts as "masculine" or "feminine" with a grain of salt when it comes to what the candidates are saying. But it's still interesting data to think about.
