Updated: Feb 3, 2020
Not so long ago the idea that smart technology could create tailored lessons for individual students was the stuff of science fiction. Now it’s close to reality – and new research about how learners react differently to visual cues can shine a light on where it might lead. Christopher Brooks, who presents his work at the London Festival of Learning this week, explains.
Back in 1995 the American author Neal Stephenson wrote a futuristic novel in which characters were presented with interactive books that could react to their environment and teach their owners what they needed to know to survive and develop.
Twenty years ago, this was science fiction – albeit fiction based on the cutting-edge science of its time. Now that fiction is becoming a reality: the technology is available to tailor course content to each learner individually. Software systems which now seem very familiar are able to rapidly adapt to environmental cues.
So, it’s important that we understand how learners might change their behaviour in response to certain cues – and that we look at how lessons we’ve already learned from more traditional classroom environments might help to inform this process.
For example, we know that women are more likely to sign up for technical subjects when they get subtle messages that the environment is gender-neutral rather than stereotypically male: existing studies tell us that changing a few objects in a classroom, such as a poster or a video game, can make a difference.
And now our new research shows that these types of cues can make a difference in online courses, too, albeit in a different way.
In a study which took place at Stanford a few years back, prospective students were shown into a classroom with various objects that were seen as stereotypically male: Star Wars and Star Trek posters, video game boxes, computer parts, soda cans and junk food. Then they were shown a classroom in which those objects had been replaced by more gender-neutral ones – water bottles, a coffee maker, art and nature pictures, lamps, general magazines and plants.
The researchers found that women participants in their study – who had been asked to imagine they were in their first job and had to choose which team to join – were significantly more likely to choose the team with the gender-neutral room. Men, meanwhile, were more likely to choose the stereotypically male room.
With massive online courses, or MOOCs, now offering learning opportunities to millions of potential students across the globe, we wanted to know whether this classroom experiment could be translated into this new and different environment.
We were able to make some subtle changes to the videos in an introductory data science MOOC on a platform called Coursera. This course had a large gender imbalance, and the proportion of female learners was only around one in six of the total. The course was huge – almost 100,000 learners had logged in during the first year. Women students were more likely than men to drop out.
Our experiment involved 46,652 students, of whom 9,683 were female. Men and women were placed in two groups, each of a little more than 18,400 men and just under 5,000 women. One group was shown a female course assistant in some of the course lectures and saw male data lab assistants in the background. The other was shown a male lecturer and lab assistants.
We were not able to change the gender of the main instructor, as this would have been too expensive. Yet there were significant differences between the two groups. Female students who were shown videos featuring more women engaged significantly more with their course materials – when we measured the number of items on which they clicked, women generated an average of 37.6 more clicks – raising their average clicks by 8.5 per cent from 480 to 517. For women viewing male-dominated course content, the average clicks were much lower – their average dropped to 442 clicks.
And there was a parallel effect for men viewing female course content; albeit a much smaller one. Their average clicks dropped by 2.3 per cent, from 508 to 496.
Women who saw the feminised content were also likely to post more in online forums relating to the course – though they were all less likely to participate than the men. Those who posted in discussions did so on average 6.28 times if they’d seen the female videos, compared with an average of 4.62 times if they’d seen the male ones. Again, there was an opposite but smaller effect for men – male posters who saw the female videos posted an average 5.67 times, or 6.31 times if they’d seen the male ones.
While students’ ability to stay the course wasn’t affected by these visual cues, some of their other course activities were.
So, what’s the significance of our work? Clearly it underlines the subtle ways in which we respond to cues that tell us we’re in a non-hostile environment. But it’s bigger than that.
Like Neal Stephenson, we’re now imagining a future in which students receive images tailored to their own characteristics and behaviour. So instead of focusing on producing one great video, course providers could provide a range of options to which different learners could travel.
Stephenson’s vision is becoming reality, and we believe the world of education is well placed to embrace it.
How Gender Cues in Educational Video Impact Participation and Retention is by Christopher Brooks, Joshua Gardner and Kaifeng Chen of the School of Information at the University of Michigan. It is being presented at the London Festival of Learning by Christopher Brooks.