| 摘要: | 在高度數位化的環境，數位資訊科學教育越來越重要。然而，程式新手在傳統的教學方法中面臨了高失敗率與高輟學率的問題。因此，本研究設計了包含合作學習與調節學習的翻轉教室模型以及SPOCs的課程模式。此外，通過線上互動程式開發環境與學習管理平台蒐集學生的編程行為、影片點擊行為以及同儕互評結果，將學生的學習歷程以學習分析儀表板呈現，幫助教師與學生評估自己的學習並制定新的學習策略。
本研究實驗對象為北部某國立大學三十一位研究所學生，研究工具包含Python能力測驗、程式態度量表、程式自我效能量表、自我調節學習量表、學習動機量表以及問題導向任務反思問卷。學生於課前通過SPOCs影片與問題解決任務對課程內容進行複習，課程中通過程式專案共創、同儕互評進行合作學習，課程後提供學生學習分析儀表板，讓同組學生參考儀表板內容進行反思與學習策略的制定，並於最後填寫問題導向任務反思問卷來評估自己本次課程的表現。
研究結果表明，學生學習表現、程式態度、程式自我效能、學習動機以及自我調節學習中的任務策略、尋求幫助、自我評估面向都有顯著提升。通過分群演算法也發現自我調節前測分數高的學生經常觀看儀表板且學習表現成長幅度較大，程式態度消極的學生經常在觀看影片時調整影片速率且學習表現的成長幅度較大。根據序列分析的結果也發現學習表現高的學生經常在暫停及調整影片速率後將影片看完，也會在觀看影片過程中重複確認影片重點與內容。本研究所設計的課程模式及視覺化分析系統提升了學生程式學習的效果，並找出學習表現進步幅度較大學生的學習行為。未來可進一步確認實驗設計在大規模課程的效果。
;As Information technology continues to advance, Computer Science education is increasingly important. However, programming novices were still plagued by the high failure rate of conventional teaching methods. This paper has designed a class model that combines SPOCs with a re-designed Flipped classroom module containing knowledge construction and regulated learning. Moreover, by using a web-based interactive development environment and a Learning management system to collect students’ coding log, video watching log and peer review results, this study shows students’ learning process in Learning Analytics Dashboard (LAD), helping students and teachers evaluate and improve their learning.
In this study, 31 graduate students from a national university in northern Taiwan participated in the experiment for 18 weeks. The research tools in this study contain Python ability exam, programming attitude questionnaire, programming self-efficacy questionnaire, self-regulated learning (SRL) questionnaire, learning motivation questionnaire and problem-oriented mission reflection questionnaire. Students review the course content through SPOCs videos and problem-solving tasks before class, collaborative learning through the co-creation of problem-solving projects and peer evaluation in the class, watch the learning analytics dashboard and develop new learning strategies with group members after class.
The results indicate that students’ learning performance, programming attitude, programming self-efficacy, learning motivation and task strategy, self evaluate, and help-seeking skills have been significantly improved. Results of the clustering algorithm show that students with high self-regulated learning score in pre-test check out LAD frequently and have more remarkable growth in learning performance. Students with low programming attitude scores in pre-test often change the video rate and have greater learning performance growth. Sequence analysis also indicates that students with high learning performance usually watch until the end after pausing or adjusting the rate of the video. They also repeatedly check the index of the video. The class design and visual analytics system improve students’ programming learning. Future researchers can confirm the effectiveness of the research design in the massive learning environment. |
