COVID-19 Resources & Updates Learn More

A new study on video in Canvas revealed something that seemed too good to be true: Canvas courses that included video had lower attrition rates than Canvas courses without video. Learning researchers like Richard E. Clark have strongly argued that video itself, like any medium, does not directly impact student outcomes such as grades, but in our study we discovered video having an unexpected outcome—retention!

We did a lot of homework to see if other variables could be responsible for what we saw in the data. We published a summary of our findings in this Canvas video usage infographic. The infographic didn’t leave a lot of room to dive into the details around video and attrition rates, and that’s why we called you here today. Let’s dive in!


-Course Attrition Rate: Percentage of students who dropped out of a course after it began

-Video Course: Course that regularly uses video (minimum of 10 videos in one course)

-Non-Video Course: Course that does not use any video

Video’s Connection to Attrition Rates

We focused on higher ed because K-12’s structure inhibits this type of attrition. We saw that once a higher ed course’s enrollment climbed above 100 students, video had a real impact. And as courses got larger, so did the impact. In courses with fewer than 100 students, video didn’t have a large enough effect to draw any conclusions.

Screen Shot 2016-10-24 at 4.52.47 PM.png

Time to Hypothesize

We first did our due diligence to make sure weird data wasn’t messing with our analysis. We didn’t have any outlier institutions throwing off our dataset; all our schools were posting overall attrition rates within the range of WICHE’s 2013 survey results on course attrition. We feel good about our data.

So what’s behind the connection? Could the medium of instruction actually impact student outcomes? I’ll be back with a post next week on what we learned by exploring this connection.

In the meantime, please share your thoughts in the comments section below or on Twitter (@CanvasLMS, use #ArcVideo). What do you think is causing this? Are there underlying variables we failed to measure? Data weirdness we may have missed? Or, in your experience, does it make sense that video can increase course completion rates? And if so, how? We’ll read your responses and return with a followup post next week.


Keep learning,
Jared Stein and David Dean
Strategy & Research Team