高中生運算思維測驗發展

Development of a Computational Thinking Test for High School Students

運算思維是二十一世紀學生必備的能力，各國紛紛將運算思維納入資訊科技課程的核心理念，相關研究亦成為電腦科學領域的趨勢。然而，如何有效評量運算思維一直是許多研究提及需重視的問題。目前評量的內容多依附在特定程式設計語言或知識下，易流於成就測驗。評估方式多透過分析程式內容或以自評量表與問卷的蒐集方式，前者較為耗時，後者偏向對運算思維的理解與態度，無法準確評量運算思維。因此，發展一套客觀、易實施的評量工具以評估真實的運算思維能力是重要的。本研究旨在發展高中生運算思維測驗，包含問題拆解、演算法、樣式一般化與資料表示等四項內涵。試題由八位中學資訊教師發展，研究者與三位電腦科學專家選題與編修，最後共12題。整體測驗難度接近適中（P = .63），所有試題在高低分群上達顯著差異（p < .05），顯示試題具有鑑別度。在信、效度部分，所有試題與總分之相關係數值皆達顯著水準（p < .01），表示所有題目與總測驗之間具有同質性。試題經專家發展與評估，故具專家效度。此外，本測驗結果與國際運算思維挑戰賽分數呈正相關達顯著水準，顯示具有效標關聯效度。整體而言，本研究發展之運算思維評量工具，具良好的信、效度，可供學生自我了解及教師教學設計時之參考。

Computational thinking (CT) is an essential skill for the 21st century. Research regarding the implementation of CT skills in k-12 education explores the issues about how to teach, however, it still lacks the discussion about how to assess. Additionally, existing research about CT assessment either assesses CT skills by grading students’ computational artifacts, which is subjective and time consuming, or develops the assessment restricting to specific programming languages, programming tools, or knowledge units. To develop an effective assessment for CT skills, this study establishes language/tool/knowledge-independent standards for CT in terms of the four CT concepts-decomposition, algorithm, pattern generalization and data representation. The initial 23 items were drafted by eight high school computing teachers, and then evaluated and revised by three computer scientists to produce final 12 items. The item analysis was conducted to examine the quality of individual items. The item difficulty index (P value) is .63 (medium difficulty). The t tests reveal a significant difference in each item between the high- and low-score groups ( p < .05). It shows the effectiveness of individual item on assessing students’ CT skills. Based on the reliability and validity analysis result, each item is positively correlated with total scale scores ( p < .01), indicating that there is a homogeneity between items and this scale. The verification of content validity was then conducted by computer scientists. The correlation between the proposed scale and the Bebras Challenge was also analyzed and the result shows the coefficient was positively significant (r = .254, p < .001). This proves the criterion-related validity. In conclusion, the CT-assessment instrument developed by this study has both high reliability and high validity. It could be an effective instrument to assess students’ CT skills.