本研究設計建模引導之科學程式設計教學，引導學生透過「現象描述」、「資料建模」、「流程建模」、「程式化」以及「觀察與除錯」等建模歷程逐步設計解題策略並編寫程式碼以解決科學問題。研究以準實驗研究法驗證所發展的建模引導教學模式之成效，教學實驗於大學通識課程實施，根據實驗結果發現：一、建模引導教學能幫助學生將複雜的科學現象擷取出關鍵解題元素、將解題相關的變數以及變數間的關係做較精確的描述，並將解題的步驟與流程較邏輯地陳述出來，因此整體來說，能提升學生「資料建模」與「流程建模」之品質，此亦分別對應運算思維的「資料表示」與「演算法設計」；對於低先備能力的學生來說，能同時提升「現象描述」、「資料建模」、「流程建模」之建模品質；二、建模引導教學能提升低先備能力學生之建模品質，藉以逐步發展解題策略，更精確表達資料並設計演算法，進而能撰寫程式以解決科學問題，因此能增進科學問題解決能力；三、由於建模輔助教學能逐步引導學生在複雜的科學問題情境中找到解題的脈絡，因此實驗組學生能感受「現象描述」與「流程建模」對科學問題解決的助益，亦較能清楚說明科學程式設計的優點，且能肯定程式設計在科學問題解決、科學理解與探索中的角色。

This study aims to design a modelling-based instruction for scientific programming. Through the guidance by a modelling-based learning platform, students had to follow a step-by-step modelling process, including "phenomenon description," "data modelling," "logic modelling," "coding," and "verification and debugging." This instruction is to enable students to gradually develop problem-solving strategies and write program code to solve scientific problems. A quasi-experimental research was conducted to examine the effectiveness of the modelling-based instruction on learning of scientific programming. The experimental group received modelling-guided instruction, while the control group received traditional instruction. Based on the results, the following research findings were identified: (1) Modelling-based instruction helped students build their solutions in various ways: it enabled students to extract key elements of the problem from complex scientific phenomena, provide more accurate descriptions of variables and their relationships, and represent problem-solving steps more logically. Overall, this approach enhanced the quality of students' "data modelling" and "logic modelling," corresponding to "data representation" and "algorithm design" of computational thinking, respectively. For students with lower prior ability, it also improved the quality of "phenomenon description." (2) Since the modelling-based instruction improved the modelling quality of students with lower prior knowledge, it helped them gradually develop problem-solving strategies, represent data more accurately, design algorithms, and then write code to solve scientific problems. As a result, it enhanced their ability to solve scientific problems. (3) As modelling-based instruction gradually guided students to find the strategies for problem-solving in complex scientific problem context, the experimental group students clearly perceived the benefits of "phenomenon description" and "logic modelling" in solving scientific problems. They could also articulate the advantages of scientific programming more clearly and affirm the role that programming plays in concept understanding, exploration, and problem-solving in the field of science.

STEM科際整合教學 ； 建模 ； 科學運算 ； 程式設計教學 ； 運算思維

STEM Education ； Modelling ； Computational Science ； Programming Instruction ； Computational Thinking