Hello everyone. In this article, I introduce a paper read in the English Literature Seminar.

• Paper Title: Deploying multimodal learning analytics models to explore the impact of digital distraction and peer learning on student performance.

• Journal: Computers & Education

• Volume: 190, Article 104599

• Authors: LIAO, Chen-Hsuan; WU, Jiun-Yu

• Year: 2022

• DOI: https://doi.org/10.1016/j.compedu.2022.104599

This paper delves into Problem-based Learning (PBL) in blended learning environments, specifically investigating how the use of social media affects academic performance. Utilizing Multimodal Learning Analytics (MLA), it analyzes how interference from digital devices, known as “digital distraction,” leads to decreased productivity and negative impacts on mental and emotional health, as well as how peer learning orientation and actual participation levels are related to academic performance. Fifty-one graduate students participated in the research, and data was collected from their statements on Facebook; machine learning was then used to investigate whether those statements were statistically significant.

As a specific analytical method, student Facebook data over the past several years was used as a training set, and a machine learning model predicted the degree to which current students’ statements were statistically related to previous data. Additionally, the study used machine learning models to predict how individual student characteristics, Facebook usage, peer learning orientation (self-reported tendency toward peer learning), and peer learning engagement (an objective indicator of learning engagement among peers, quantified by a machine learning classification model that categorizes discussion messages in private Facebook learning communities into statistically relevant or irrelevant messages) affect academic performance. This analysis showed that higher levels of active engagement in peer learning correlated with better academic performance, confirming that social media is effective as a site for social learning. Through the utilization of rich data including surveys and questionnaires, this research clarified how specific educational elements influence academic performance. Specifically, it objectively and accurately measured learners’ peer learning engagement in a PBL environment and showed that its predictive validity for blended PBL learning performance was stronger than that of subjective peer learning orientation.

Furthermore, this study focuses on the negative effects of digital distraction on academic performance. By evaluating the extent to which students are exposed to digital temptations and how that affects their grades, it provides suggestions for how educators should address these challenges. It argues for the importance of strategies to reduce digital distraction and promote peer learning within the context of blended learning. The findings obtained from this research provide hints for designing more effective educational strategies and interventions in the application of educational technology.

The following are my thoughts.

This research adopts a multimodal approach to explore a wide range of factors—including peer learning, social software usage, distraction, and individual differences among learners—to predict the academic performance of learners engaged in problem-based learning within a blended learning environment. This approach combines multiple variables from different perspectives to explore how these factors influence learner performance. Specifically, it analyzes how the degree of active participation in peer learning and the way social media tools are utilized are reflected in academic performance.

Furthermore, this paper demonstrates the specific impact of these factors on academic performance alongside actual data. Learners with a higher orientation and degree of engagement in peer learning generally tend to achieve better academic results. Additionally, by utilizing multimodal learning analytics and machine learning techniques, the study closely examines the impact of different educational factors on academic performance. This method makes it possible to clarify even minute influences that were often overlooked in conventional research, and new applications in the field of education are expected.

However, I have several questions. This paper uses MLA to multi-dimensionally analyze the impact of various PBL elements on academic performance in a blended learning environment. However, descriptions of what kind of PBL tasks were actually set and the specific mechanisms of the machine learning models are insufficient, leading to a lack of clarity in those respects. Since these factors strongly influence the results, I feel that the validity of the results cannot be judged unless these points are clearly explained. In addition, I believe a detailed analysis of which PBL elements were particularly influential is necessary. Furthermore, Facebook posts made by students over the past several years were used as a dataset. Using this data, machine learning was used to predict the statistical relevance of current students’ statements and compared with artificially performed coding results. I believe that the purpose of this process and its educational significance have not been sufficiently explained. While it was an interesting paper, I also felt there were many unclear points.

While the practical application of MLA in educational settings is still in its early stages, I am interested in exploring the impact of interactions between different elements on academic performance. The appeal of MLA lies in its use of various datasets that can represent specific characteristics in greater detail. I hope to deepen my knowledge by reading further research papers in this field in the future.

(Written by: Li Tang, 2nd-year Master’s student)