Men dominate the news world the same as they do everything else, which is to say, most news stories are written by men about men, and most news photographs depict men. Now, researchers have found some of the strongest evidence of that one-sidedness yet, along with a curious twist: Across nearly all topics and outlets, the likelihood of seeing women in photos is higher than the likelihood of reading about them and their opinions.
Those findings support a longstanding view that we continue to value women—apparently, even the women who editors deem newsworthy—more for their appearance than their intellect, write Sen Jia and his colleagues at the University of Bristol and Cardiff University.
Though the findings themselves may not be all that surprising, Jia and his team reached their conclusions using a novel approach and a truly gigantic data set. Over six months, the researchers collected 2,353,652 news articles from 950 online news sites around the world, nearly 13,000 a day.
The likelihood of seeing women in photos is higher than the likelihood of reading about them and their opinions.
No, the team did not read all of those articles. Four of the five researchers are computer scientists—they had a computer do it.
Specifically, they fed news stories and accompanying images to several classification algorithms. One algorithm was built to identify the topics described in each news article, another to identify the sex of people mentioned in the articles, and a third to find faces in photos and identify their gender. Although they applied to different categories, each algorithm had a similar structure: They learned different categories based on a sample set. For example, the topic classifier learned the differences between stories about politics, entertainment, and so on by reading, so to speak, the corresponding sections of the New York Times and Reuters.
Unsurprisingly, men were overrepresented, though to varying degrees. Across publications and topics, just 23 percent of the people mentioned in a story were men, with women best represented in the entertainment section (about 30 percent of those mentioned were women) and worst represented in sports (about five percent).
The story is similar in photos, except that women were somewhat better represented. Overall, 30 percent of the people in photos were women. That number goes up to about 40 percent in the entertainment section, but drops to 25 percent in sports.
The most balanced topic—in the sense that women were equally underrepresented in both text and photos—was politics, where roughly one-fifth of the people mentioned in articles and shown in pictures were women.
None of this is good, Jia and his colleagues write, but their approach may help make for a better future. By developing automatic methods for detecting gender bias in the news, the team writes, they’re able to analyze very large amounts of data quickly and in great detail. “As such, they offer hope in efforts to challenge the marginalisation of women’s voices in the news media and in so doing acknowledge the value of their contributions to the future strength of deliberative democracy.”
Quick Studies is an award-winning series that sheds light on new research and discoveries that change the way we look at the world.
