Abstract:

In the last weeks of the pandemic crisis provoked by the Corona Virus much has changed in education institutions, namely in the Higher Education Institutions (HEI). The need to create “social distance”, meaning a physical separation between people to prevent the virus from spreading, completely changed the traditional paradigm of lecturing, monitoring and assisting. The concept of a classroom, or of a laboratory, just do not apply anymore.

One of the techniques that current technology offers educators to circumvent the problem is the possibility to pre-record video lectures, which can be saved in high resolution, and be automatically indexed, and even gathered with other video streams (different video sources) in a single file, or link. In theory, students have access to this kind of pedagogical material, which never in history could have so much quality for something created at home. Fulfilling premises of e-Learning, students may visualize it according to their learning pace, at the time and place it suits best to each one. It may seem these tools were just waiting for a good opportunity to be used.

The reality, however, is much less rosy than that.
In this study, we used two courses: one from a first-year at university level, and another from a third-year university level, which is also given (as shared) for a masters’ course. The datasets allow us to compare behavior from “freshmen” to final year students, and to post-graduate students. In total, we had access to more than 200 students. The data for our research is a series of pre-recorded videos in Panopto in each course, and the statistics that Panopto automatically provides from the student visualizations of those sessions. All datasets were anonymized before the start of our research.

The statistics offer a series of Comma Separated Values (CSV) files, which were further and thoroughly analyzed by us. The information conveyed in the files is about the number of accesses, time of access, and points of the restart, i.e., in which part of the video file the student presses the playing button.
We can say, with guarantee, that students most of the time do not see the video in a straight way from the beginning to the end. Rather, the “scan” the video for parts. Most of the time, they play for one minute, then advance the video for 5 minutes and play for just 15 seconds, then they go the end, and then back to some point in the middle and continue this random process until they feel the video is all checked.

Panopto reports that the rate of completely seeing all parts of the video is around 90%, sometimes 60% and sometimes less than 20%. Therefore, our research leads us to the question: do the students actually see what is most important in the video?

To assist us in this research we create an encoding of the viewing pattern for each student. This representation allows us to compare and aggregate visualization patterns which ultimately allowed us to resonate about the whole class pattern. Therefore, we are able to compare patterns from the different types of students (first years, last years and post-grad students).

Results prove that there is a substantial difference between the groups and that experience/maturity plays a very important role when using this methodology to convey lectures at a distance.

Keywords:

Video lectures, self-study, e-learning, student behavior.