Abstract:

Although temporality is embodied in instructional design, implicitly present in several learning theories and central to the self-regulation of learning and awarding of credits, it is poorly studied in the field education. Furthermore, the construct of time is poorly defined and at times misunderstood. A possible remedy is to take into account all the levels at which time exists and plays a pivotal role in learning. In addition to the commonly studied micro level (time-on-task), we need to focus on meso level (curricula) and macro levels (organization of courses, credits, and certification).

This learning analytics study included four higher education courses in medical education over a full year duration. All the courses followed a problem-based learning approach. Each week, students have to work collaboratively on a new online problem, typically in small groups. Students were classified as high and low achievers based on their performance. Temporality was studied on daily, weekly, course-wise and year wise. The patterns of each group (low and high achievers) in each period were visually plotted and compared. Correlation with the performance was performed with the non-parametric Spearman correlation coefficient using re-sampling permutation technique.

On the day level, the patterns of interactions were almost identical for high or low achievers and were rather expected. Interactions were at the peak after working hours, and trough during late night. On the week level, there was a distinct pattern of high achievers always contributing early during the week. The pattern was consistent in every course and throughout the year. The number of early bird interactions was significantly and consistently correlated with better performance. On the course level, high achievers tended to participate earlier, and low achievers tended to participate late. However, it was not statistically correlated with performance. Visual examination of the patterns of interactions revealed that a group of low achievers start the course motivated to participate. Nonetheless, this group loses their motivation early. This might be an indication that support might have helped this special group of students to sustain their efforts. On the year level, high achievers contributed early, as the academic year advanced, low achievers were almost comparable to high achievers in terms of activity. By the end of the year and close to the exams, low achievers were significantly outperforming the high achievers in terms of activity. The number of interactions was only correlated with performance early during the first course of the year.

The findings of this study highlight some important points, temporality is a defining factor of how students regulate their learning and should be taken into account when designing a monitoring system. High achievers are always active early in the year, and on assignments. Low achievers, on the other hand, tend to be significantly more active close to examination times. It is reasonable here to conclude that performance predictors are not constant and they change according to the course organization and the exam time. Learning analytics systems and dashboard designers need to be aware of this issue and design a temporality aware applications.

Keywords:

Learning analytics, temporality, collaborative learning, time, procrastination.