December: video project update
The Methods 3 videos have accrued more than 11,000 minutes of views: an average of 37 views each with an average view duration of just over 10 minutes.
One can make some interesting deductions about study patterns. Methods 3, Lecture 10, part 1 (with 92 views at an average of 14.41 minutes per view) is the most watched. It concerns the theory and computation of Fourier series and the application of this to the solution of the heat equation. A large number of these views occurred the day before a homework was due which made heavy use of the material in this lecture. As a counterpoint to this, reading week (when the students had extra time to catch up on lectures) was a time of relatively low-use. In general, activity peaks around the time when homework deadlines loom and I anticipate that we will have another huge peak around exam term.
I have had some very positive feedback about how useful the videos have been (even offers to film for free!) and some excellent suggestions for improvement. These included adding the date of the lecture to the video title (for ease of comparison with lecture notes) and giving more detailed descriptions of lecture content in the video caption (including timings of different sections of each lecture); this would make it easier to navigate through a solid hour of material. While I don't have time to do this scrupulously, I have added some pointers in some of the videos.
The audio quality is still pretty poor and I do not know how to improve this.
A Moodle feedback questionnaire had relatively low uptake (11/195 responses), so nothing of statistically significance can be said. All but one respondent thought videos were useful in principle, but by its nature the questionnaire is likely to attract responses from students who were most enthusiastic about the videos in the first place. The comments left were extremely positive.
I had to compress my videos much further before they were able to be uploaded to Lecturecast, and the Lecturecast system has proved very user-unfriendly. For instance, uploads are made using a Java applet which fails to run on Linux because it requires unrestricted permissions (even Windows warns you about this and explains that the code will soon be deprecated). What’s wrong with a simple HTML upload form?
Note that eight out of the eleven students who responded to the Moodle questionnaire preferred the lectures to be available on Youtube rather than Lecturecast. Two had no strong opinion. One preferred Lecturecast. I do not see that the sample bias present in the questionnaire (self-selecting the most enthusiastic students) would correlate with this particular variable, so I am inclined to take this at face value (bearing in mind that the sample size is rather small).
The next step
At this point in the project two things are clear:
- Some students are making good use of these resources and appreciate their availability.
- The scheme would be difficult to implement on a large scale without
significant help from administrative and technical support staff.
- Each hour of video is currently taking four hours to compress and further time to upload. My laptop is running late into the night and at the weekend. This would need to be automated and done on a UCL-based computer.
- We would need more cameras and more SD cards (to avoid problems with batteries running down and memory running out). And better quality microphones.
- The cost of hiring students to film is nontrivial. I think that
having mathematically competent human being filming has made a
significant difference to these videos and we have much better
quality as a result. But is there room in the department budget to
run such a project again and on a larger scale?