The stereotype that women and tech don't go together is alive and well, according to a new study that analyzed how Twitter users guessed the gender behind anonymous tweets. When women tweeted about tech, or anything serious, really, they were often mistaken for men.
With the help of natural language processing (NLP), a form of artificial intelligence used in linguistics and computer science, a group of researchers from the University of Pennsylvania Positive Psychology Center, Germany, and Australia looked at the way language influences stereotypes online. They published their findings this month in the journal Social Psychology and Personality Science.
The researchers found that while more than 400 readers accurately guessed the gender of a tweet's author, they erred on two fronts: When women tweeted about tech, business, or the news, readers most often thought they were men. Meanwhile, men who Tweeted about "positive sociality," such as having fun with family and friends, were most often mistaken for women.
Readers also thought a woman was a man if she had a "wide and serious perspective on things,” including tweets about the news or matters that are important outside the personal realm, lead researcher Jordan Carpenter told Motherboard. Words like “history”, “research” and “war” were automatically associated with men, even though they were tweeted by women.
"So the conclusion that I'm drawing is one that ties into the well-known set of stereotypes against women in STEM, tech, and science," Carpenter said.
Readers also thought a woman was a man if she had a "wide and serious perspective on things.”
The stereotypes extended into politics. When readers were asked to identify a Tweet author's political orientation, they most often assumed "feminine-sounding people were liberal and masculine-sounding people were conservative," he said.
This is not unfounded. Even in our most recent election, women did vote for liberal candidates more than men, according to the Pew Research Center. And these findings echo the theory by linguist George Lakoff, who has theorized how cognitive metaphors influence thought.
For example, the "nurturant parent [mother]" metaphor is most often associated with liberals, while the "strict father" metaphor is most often associated with conservatives. Carpenter said people tend to associate compassion and care with liberals and women, while law and order are most often associated with conservatives and men.
"Within psychology there's this big debate about the appropriateness of using stereotypes," he said. Some believe that stereotypes are harmful and usually false, though others argue that they're mostly true and okay to use. "What this study tries to do is add nuance to that."
With the help of the NLP technique, Carpenter's group wanted to focus on not just language, but which traits in people are represented in their words. They’re hoping the research will help isolate inaccurate and accurate stereotypes, and defend the idea that stereotypes can be useful unless they’re false.
"We shouldn't stereotype stereotypes. Stereotypes are more useful than not with the big asterisk: The ones that aren't have nothing to defend on. They're socially harmful, people use them badly, and are not using them to draw accurate conclusions."
Get six of our favorite Motherboard stories every day by signing up for our newsletter.