In recent years, there has been a growing interest in learning analytics and educational data mining in the higher education sector. Learning analytics data can be used to identify at-risk students and to help instructors identify how students are engaging with their online course materials.
Despite the popularity of video-based instruction in higher education, there is limited research to-date on how instructors can use analytics data to investigate video viewing patterns, with a view to determining the efficacy of those videos. Analysing video-watching patterns provides a unique opportunity to appreciate how, and if, students learn more effectively via video. To that end, this case study explores the video viewing patterns of a cohort of 348 undergraduate business students taking a business-oriented IT module. The students had access to a series of 17 videos, spanning five practical Microsoft Excel topics, which were developed specifically for a module entitled ‘Business Information Management’. Students attended two one-hour lectures per week and five one-hour computer labs over the semester. In addition to an end-of-term theory exam, there was also a one-hour end-of-term practical spreadsheet exam.
This case study answers the following questions:
• To what extent do students use instructional videos as a tool for initial learning and revision for the end of term practical exam?
• Does the difficulty of the material affect video viewing patterns?
• How much [what proportion] of the videos are watched?
• Does the difficulty of the material affect how much [what proportion] of the videos are watched?
• To what extent do students watch a series of videos on a topic?
The paper demonstrates the nature of data that can be freely obtained from YouTube analytics and how it can be further exploited to determine how instructional videos are being used (how many students access the videos, for how long, and when). The paper also highlights the importance of undertaking a deeper analysis of the data, as the initial summary data may be misleading.