Online education was once hailed as a potential equalizer, offering science, technology, engineering and math skills to everyone regardless of gender, nationality or socioeconomic status. Online, students could avoid stigmas, such as being one of the only women in a class.

It didn’t turn out that way.

“Very quickly we saw that’s not the case,” said Rene Kizilcec, assistant professor of information science. “Many of the same barriers we see in in-person environments do replicate online, and they get amplified, because so many more people have access to these platforms.”

In new research, Kizilcec found that adding a photo of women and an inclusivity statement to a Facebook ad for a computer science course increased the number of women who clicked on the ad by 26 percent. Similar changes to an enrollment page raised the number of women who signed up for the course by up to 18 percent.

The findings suggest that relatively simple changes can boost underrepresented groups’ participation in online STEM education – a critical area, as one in three U.S. college students takes at least one online course and 15 percent are enrolled in entirely online programs.

“The cues that we place in digital environments are powerful shapers of participation,” said Kizilcec, lead author of “Psychologically Inclusive Design: Cues Impact Women’s Participation in STEM Education,” which he will be present at the Conference on Human Factors in Computing Systems, May 4-9 in Glasgow, Scotland.

“Who enrolls in a course is in part determined by the first impression people form and that is shaped by what they see and read early on,” he said. “For designers and creators of content, that is important to keep in mind.”

Kizilcec and co-author Andrew Saltarelli of Stanford University conducted two experiments with the Stanford Center for Professional Development. In the first, they created a marketing campaign on Facebook showing four versions of an ad to around 200,000 people expected to be interested in online computer science classes. One version displayed a generic image of a computer with no inclusivity statement; one showed a photo of women with no statement; one showed the image of the computer with text including, “The history of computer programming is a history of WOMEN”; and the fourth showed both the photo of women and the inclusive wording.

The ad with the inclusive text and photo significantly increased the number of women who clicked by 26 percent over the default version, which contained the generic computer image and no inclusivity text. For three of the versions, men’s engagement was unaffected, but fewer men clicked on the ad with the picture of the computer and the inclusivity text, perhaps because they were confused by the ad’s objectives, Kizilcec said.

“The text alone without the context of the image may have violated expectations,” he said. “It was a little out of context for them. It highlights the importance of combining the visual and the textual in ways that make sense as a whole.”

In the second experiment, the researchers manipulated a STEM course’s enrollment page to alternately include a photo of women and a statement including the words, “Anyone can enroll from anywhere in the world and everyone, no matter what their gender, ethnicity, or socio-economic status, can be successful.” They also showed a version without the inclusive text and photo.

Two out of three times, more women enrolled when the page showed the inclusive photo and text, but one-third of the time it had the reverse effect, and fewer women signed up for the course. The researchers plan to explore this result in future work, but hypothesized based on subsequent research that it could have to do with the women’s ages.

“Something about what generation you’re from could affect how you interpret these cues,” Kizilcec said.

The study casts light on how online environments could be altered or customized to increase diversity, and how women respond to these digital tweaks. Prior work by other researchers has shown that manipulating classroom design – replacing Star Trek posters with generic images of nature, for instance – can significantly increase women’s intentions of signing up for STEM classes. But until now, no one had measured whether women actually follow through on those intentions.

“There is a gap between intention and behavior,” Kizilcec said. “Our goal with this work is to better understand what people actually do when you change these cues.”

This article, written by Melanie Lefkowitz, originally appeared on the Cornell Chronicle on March 20, 2019.