Want Lower Drop Out Rates? Use Video: Part 2

In last week’s Canvas blog post, our research identified a connection between video use in large higher ed courses and course completion rates. (If you haven’t read that post yet, we suggest you pause and go read it here). This week, we’ll show you what we learned about social connectedness and student engagement as we explored that connection.

Course Type

We first questioned whether video’s impact on attrition was specific to fully online or blended courses. Our super smart data scientists created a machine learning model to divide higher ed courses between these two course types. We focused on average-sized (10-50 students) and large (150+ students) courses.

Screen Shot 2016-11-04 at 10.08.07 AM.png


In the average-sized courses, the effect was the same regardless of course type: negligible (d = 0.11 for both). With larger classes, video had a medium impact on blended courses (d = 0.64) and small impact on fully online courses (d = 0.23).

Social Connectedness

Both course types experienced the same effect, so we decided to look at social connectedness as a possible explanation. In small classes, students have many opportunities to connect with their teachers and other students face-to-face. However, as classes grow, opportunities to connect can shrink. Borup, West, and Graham (2012) found that video may increase the sense of humanness and decrease the sense of distance online. "Many students [felt] that video-based communication made their instructors seem more real, present, and familiar...". Maybe large classes can use video to compensate for their seemingly impersonal nature. But does video alone accomplish a sense of social connectedness? Or do all LMS activities do this?

Instructor Effort = Good. Video = Good.

We ran a linear regression model on our large-course dataset to see if LMS activity in general increased social connectedness, and we found that instructors who used video tended to use more LMS activities overall. And their courses had the lower attrition rates. So yeah, instructor effort matters (duh).

But then we controlled for all those other LMS activities to focus on video’s effect. When controlling those variables, our model still placed the attrition rate for video courses 5.8% lower than the attrition rate for non-video courses (p < .001). Which means yes, video itself does have an impact.


Recent studies show there’s still progress to make in the area of student engagement. WICHE’s 2013 survey found attrition rates of 19% for face-to-face courses and 22% for online courses. While this represents improvement over results for online courses from earlier studies, the problem remains.

At Instructure, we want to create technology that increases not only student access, but also student success. Comparing video and dropout rates uncovered an opportunity to do just that. Our analysis backed up the known importance of the human factor—instructors’ efforts to build engaging courses have a real impact on attrition. And while many LMS activities correlated to higher course-completion rates, video’s correlation was particularly strong. In short, when it comes to building the kinds of courses that keep students engaged in class, video is an especially effective medium.  

To learn more about video in the LMS, go check out our infographic on the topic.

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