大規模開放線上課程學習分析促進科技學科教學知識之研究

Applying Learning Analytics to Enhance the Technological Pedagogical Content Knowledge of Teachers Teaching Massive Open Online Courses

過去幾年，大規模開放線上課程（MOOC）已在全球蓬勃發展。本研究主要是針對研究者在校內通識開設自行研發的「大數據的設計思考」MOOC，進行數位學習實踐研究，以增進教師個人的科技學科教學知識，並探討MOOC期末通過率的問題。研究對象為研究者107學年第一學期開課之班級306位大學生，蒐集其在修課平臺中學習歷程共187萬1,747筆資料，進行學習分析，探討學生背景變項（性別、學院與年級），以及觀看教學影片的完成度、運用自我評量與觀看教材的策略、額外搭配多元載具等三項學習行為，是否影響學生期末修課結果。統計處理採卡方檢定和邏輯斯迴歸模型。研究結果發現：學院別和前述三項學習行為，對於學生期末通過課程及格標準皆有顯著影響。同時利用和諧性分數和ROC曲線，檢驗的學習成效模型，和諧性分數達83.8%，顯示具有高預測率。本研究結果未來可應用於發展學習預警機制，以提高學生學習表現與修課通過率。 Over the past few years, massive open online courses (MOOCs) have flourished globally. This study investigated big data-based general education MOOCs developed by its author to learn about e-learning in practice, enhance the technological pedagogical content knowledge of teachers, and determine students' MOOC passing rates at the end of a semester. The study participants were students who studied in the aforementioned MOOCs in the first semester of academic year 2018-2019. Learning analytics were performed on the learning process of 306 students (which contained 1,871,747 pieces of data) to identify the students' background variables (i.e., gender, college, and grade level) and determine whether the three “learning behavior” of students (i.e., how much of the educational videos they had finished watching, self-assessment and textbook-viewing strategies that they had adopted, and the diversity of devices that they had used to support their learning) had an effect on their MOOC results at the end of a semester. A chi-square test was conducted and a logistic regression model was used to perform a statistical analysis, where the results showed that college and the three learning behaviors exhibited a pronounced effect on whether the students passed the MOOCs at the end of a semester as well as their passing rates. In addition, this study used harmony scores and ROC curves to test the effectiveness of the student learning model. The results showed a harmony score of 83.2%, signifying high accuracy. In the future, early-warning systems can be developed to elevate students' academic performance and MOOC passing rates.