Estimating the cognitive value of YouTube's educational videos: A learning analytics approach

Abstract

The literature provides some evidence for the effectiveness of online educational videos in students' learning. Also, general features of educational videos such as production style and video length were investigated to some extent. However, the actual cognitive features of educational videos available on the Internet have remained almost untreated, so far. The main goal of this study is to use learning analytics to investigate how far educational videos on YouTube support cognitive features –as formulated in the cognitive theory of multimedia learning– and how far these features support students' learning. We measure the Video Cognitive Value (VCV) by viewers' rating, particularly using the number of Likes based on a survey that we conducted to understand the semantics of Likes and Dislikes of YouTube's educational videos. Then, a sample of 105 videos was collected and analyzed with respect to cognitive features. A regression analysis showed that only four out of ten investigated features are significant for VCV (pretraining, modality, spatial contiguity, and embodiment) and the regression model could only explain 63% of the data variance. Further tests were performed to include other factors such as the video production style, the video length, the talking speed, the gender of the speaker, and whether she/he speaks English as native language. The expanded model showed only a slightly improved adjusted R-square value (68%). These results suggest that further research is required to identify and specify additional cognitive features in educational videos towards deeper understanding of the video cognitive value.