A few months ago I blogged about my summer 2013 research on optimal video length for student engagement. Recently Juho Kim, a Ph.D. student at MIT, Rob Rubin, the VP of Engineering at edX, and I extended that work and published How Video Production Affects Student Engagement: An Empirical Study of MOOC Videos at the ACM Conference on Learning at Scale (L@S). Here’s a brief summary of the paper and what we discovered.
Videos are central to the student learning experience in many kinds of MOOCs.There are four main kinds of videos on the MOOC platform:
a.) a recorded classroom lecture,
b.) an instructor’s talking head,
c.) a Khan-style digital tablet drawing (popularized by Khan Academy),
d.) a PowerPoint slideshow.
Question: How does video production affect student engagement in MOOCs?
How we went about it: We measured engagement by how long students watched each video and also whether they attempted to answer post-video assessment problems.
We took all 862 videos from four edX courses offered in Fall 2012 and hand-classified each one based on its type (e.g., traditional lecture, problem-solving tutorial) and production style (e.g., PowerPoint slides, Khan-style tablet drawing, talking head). We automatically extracted other features such as length and speaking rate (words per minute). We then mined the edX server logs to obtain over 6.9 million video watching sessions from almost 128,000 students.
To our knowledge, this is the largest-scale study of video engagement to date.