Matt Rourke, Associated Press
This April 26, 2017, photo shows the Twitter app on a mobile phone in Philadelphia. A new website allows people to upload their Twitter archives and tells them if they're more likely to retweet men or women.

SALT LAKE CITY — Not all forms of sexism on the internet are obvious. Something as simple as who you choose to follow and retweet on Twitter can have the effect of amplifying or silencing women's voices, according to open-data researcher Bastian Greshake Tzovaras.

That's why he created a tool called TwArxiv that allows Twitter users to check their unconscious biases and see how often they retweet or reply to women versus men.

"To whom are you paying attention? Is it the always same male crowd?" Tzovaras, director of research for Open Humans and co-founder of openSNP, a project that puts crowdsourced personal genetics data into the public domain, wrote in an article for Scientific American. "Make sure to identify those voices you’ve ignored so far and listen to them."
Data from an unknown Twitter user shows the user replies to more men than women but also shows that disparity has decreased over time. A new online tool called TwArxiv lets users evaluate their Twitter use for evidence of unconscious gender bias.

Tzovaras and others argue that professional fields, including those related to science, technology, engineering and math, benefit from the diverse opinions brought by women. Author Rebecca Solnit goes further in an essay for The Guardian, arguing that amplifying men's voices over women's dehumanizes women by rendering them voiceless.

Research suggests male voices are often favored on Twitter.

A 2018 study found that male political reporters in the U.S. retweeted other men three times more often than they retweeted female journalists. (The study included 2,292 journalists with a public Twitter account who were credentialed to cover Congress for an English-language outlet, of which 57 percent were male.)

In 2013, an app called Twee-Q showed, based on Twitter data from more than 30,000 users, that men were retweeted almost twice as often as women. And in 2009, Harvard Business School researchers found that men were almost twice as likely to follow a man as a woman. Women were also inclined to follow more men, but to a lesser degree — they were 25 percent more likely to follow a man than a woman, the study revealed.

"Twitter can be a highly gendered experience and there is plenty of research showing that such biases overwhelmingly favor men," the TwArxiv website reads.

But it could just be a matter of who's posting on Twitter in the first place.

While slightly more women than men use Twitter in the U.S., according to a 2018 Pew Research Center poll, a Statista survey revealed that worldwide, 35 percent of Twitter users were female and 65 percent were male as of January 2019.

Tzovaras himself shared more tweets by men than women before he consciously made an effort to be more equitable, he said on Twitter. His personal data shows he retweeted more men until about 2016.

"On Twitter, a good indicator of amplification are retweets," the TwArxiv website reads. "These can be gender balanced or show biases."
Data from an unknown Twitter user shows the user retweets posts from more men than women. A new online tool called TwArxiv lets users evaluate their Twitter use for evidence of unconscious gender bias.

Another consideration is that many industries are male-dominated to begin with. Quartz science reporter Olivia Goldhill said she was worried her Twitter data would show a bias because she focuses on the male-dominated fields of philosophy and brain science. She was surprised to see her retweets and replies were mostly equal for men and women, she wrote.

"Of course, there are several factors that could contribute to a gender imbalance — perhaps you have more male friends, or colleagues, or just retweet and reply to everything that former President Obama says," wrote Goldhill. "The data simply offer a chance to reconsider those interactions, and question whether there’s anything you can do differently."

The pattern of male voices being disproportionately amplified holds true in other mediums as well. Fewer op-eds are written by females than males and reporters are beginning to examine the gender of people they quote most often. Even in real life interactions, women tend to be interrupted more often than men and speak significantly less in professional meetings.

Other tools, such as Twitterlytic, allow users to find out the gender breakdown of followers and people they are following.

" Maybe the single best, most actionable thing is this: step back, shut up, give women space, and listen to them. "
Data researcher Bastian Greshake Tzovaras

For TwArxiv, the gender classifications are predictions based on users' first names as given in their Twitter accounts. Unknown classifications are ignored, according to the website.

"Ideally these graphs would include non-binary folks. Doing this is a bit trickier. It is thus a work in progress," the website reads.

As of the first quarter of 2018, Twitter had 336 million monthly active users, making it the fifth most popular social network among U.S. adults.

"Even something as seemingly small as retweet can, over millions of users and interactions, build into something more significant," wrote Goldhill.

16 comments on this story

Tzovaras tweeted that so far, more than 300 people have uploaded their Twitter archives to the TwArxiv website. About 65 have shared their results, allowing the public to view data visualizations representing their Twitter habits.

"Ask to whom you are giving an audience," Tzovaras advises in his Scientific American article. "Make sure also to boost the messages of women instead of only focusing on your (male) buddies."

"Maybe the single best, most actionable thing is this: step back, shut up, give women space, and listen to them," he wrote.