What is the optimal length of an instructional video?

I will do this post in English, so that the person that pointed out this short post by Philip Guo doesn’t have to resort to Google Translate to be able to read it. I at first wanted to just post it as a comment on Seb’s blog, but wanted the additional space I could use if I did it as a separate blogpost.

The post by Philip Guo briefly talks about research he is doing on the videos that are being used in EdX MOOCs. He makes some interesting assumptions which may become clearer when he posts more about his research, but which so far had me disagreeing with him profoundly.

First his conclusion: “The optimal video length is 6 minutes or shorter”. And like Seb, I am having my doubts with regards to that conclusion.

Again, some of the doubt may be lifted when Philips reports more about his research. And he for example explains this assertion: “Traditional in-person lectures usually last an hour, but students have much shorter attention spans when watching educational videos online”. Now, I know that is stated often that people have shorter attention spans when watching video on YouTube, some even say “less than 2 minutes is best“, others say “2-7 minutes” or “3-7 minutes” or just “less than 10 minutes”. These numbers can be found elsewhere but research supporting those numbers more often than not is lacking in those articles. And even so, comparing a YouTube video about cute cats with the videos that are part of a MOOC, is comparing two completely different things from a point of view of why you would want to watch more or less of a video.

For example, I am participating in a Coursera MOOC about Statistics (see screenshot) and all of its videos are longer that the mentioned 6 minutes. I dare state that I have watched all of them so far and all of them in full. Albeit I have watched most of the ones in the first 5 weeks at 1.5x regular speed after I downloaded the videos (which means my views are not tracked by the system). The lab videos showing how to do things in R, I watched at regular speed and sometimes, parts of them, I watched again.
Am I such a special student?

Then again, what numbers did Philip compare? For example, during my PhD research into the use of recorded lectures by students, we removed all the really short sessions because we assumed those were the students just checking out the recordings instead of actually engaging with them. Did Philip include those use-cases? I think there are a number of other analyses that Philip could do: for example, within the same course, comparing longer and shorter videos. Also, “bite-sized pieces” as suggested by Philip does not only refer to length, but also to a video addressing one topic or multiple topics. We would need to know more about the type of video’s that are being compared. Quality of the video could also be of influence, the ones in the Coursera MOOC that I talked about, are of high quality, others are of less quality. Could that also count in part for the viewing length.

I hope to read more about Philip’s research in the coming months. I do not think that the length of a video itself is the most important factor. But analyzing viewing behavior as a way to learn more about what students do with the videos is a good first step.