Personalized, Adaptive Interventions in MOOCs

Dan develops methods to gain a deeper understanding about how the design of online learning environments affects learner success and engagement, often by implementing and testing instructional interventions at scale. In the future Dan and his colleagues will aim to increase the adaptivity and personalization capabilities in MOOCs in order to better cater to each learner’s individual needs and preferences.

Daniel Davis is doing his PhD at the Web Information Systems Group of Delft University of Technology on the effect of various instructional interventions on learner success and self-regulation in Massive Open Online Courses (MOOCs).Together with two other PhDs of the Leiden-Delft-Erasmus Centre for Education and Learning (CEL) he wants to uncover the success factors of open online education. 

How did you become so interested in online learning?

‘While I was pursuing my master’s degree in Communication, Culture and Technology at Georgetown University, they were just making their first MOOCs. I recorded all the videos for the Bioethics MOOC. I was really interested and thought it was cool, so I stayed involved. When I learned that MOOCs will generate data on how learners engage with all the videos I recorded and all the course content I helped design, I wanted to dig into it and explore. That was the basis of my master’s thesis, which formed the path into my PhD. In Delft I could continue into this field of research.’

And what is your field of research?

‘Designing online learning environments to increase learner engagement and success. My first paper in Delft was on learning paths in edX MOOCs. EdX, the MOOC platform TU Delft works with,  provides students with a specific path to get through the course. To be successful in a MOOC you don’t necessarily have to follow that path but it’s recommended to get the most out of the course. I wondered to what extent students follow that line.

Using existing literature and our own observations we looked at two things. Firstly, how students behave without any alterations to the design. Do they deviate from the learning path and if they do, how? Secondly, whether we can design interventions to nudge the students in the right direction. The biggest challenge for students are their self-regulation skills. How do you spend your time and resources to commit to the course? In a classroom situation you can look at your peers to see what is needed to complete a course. Online you don’t have that.’

Are you using experiments in your research?

‘Definitely. Since that first paper we did nine experiments on live MOOCs. We used A-/B-testing to measure the differences between groups: one group would get the intervention, the other group wouldn’t. Right now we are evaluating the effects of the interventions on two things. Firstly on overall engagement, meaning how active students were in the course. And secondly on completion rates. But this is very basic. Every following study will go beyond this.’

And you designed all the interventions in the experiments yourself?

‘I did. I wasn’t completely new to designing and programming but the last two years I had to learn a lot of new programming languages. First Python and R, the last six months mainly Java Script, HTML and CSS. I spent this winter in Massachusetts so I could focus fully on these new languages alongside the experts at Harvard, MIT, and edX.

What effect could your research have on future MOOCs?

‘We are now also taking the intentions of students into consideration when designing the interventions. Some learners want to follow all videos and assignments, some want to get through the course as quick as possible. There are even learners that are not interested in finishing the course at all but only want to do certain parts. This means that I have to design personalized interventions. Learners that only want to do specific parts of the course will get personalized links to get to the right part. This reduces the amount of wandering through the course.

Another intervention I have designed is aimed at giving students feedback on their own learning behaviour. It’s a spider chart that shows two layers. The first layer is how students that successfully finished the course have spent their time and resources. The second layer is how you spend your own time and resources. That way you can compare the study behaviour of successful students with your own study behaviour. We hope that his will help students to reflect on their own learning.’