行政院國家科學委員會專題研究計畫 成果報告
數位實驗課程對高中生學習科學的影響(第 3 年) 研究成果報告(完整版)
計 畫 類 別 ： 個別型
計 畫 編 號 ： NSC 95-2511-S-011-001-MY3
執 行 期 間 ： 97 年 08 月 01 日至 99 年 01 月 31 日 執 行 單 位 ： 國立臺灣科技大學技術與職業教育研究所
計 畫 主 持 人 ： 陳素芬 共 同 主 持 人 ： 陳秀玲
報 告 附 件 ： 出席國際會議研究心得報告及發表論文
處 理 方 式 ： 本計畫涉及專利或其他智慧財產權，2 年後可公開查詢
中 華 民 國 99 年 03 月 31 日
1
行政院國家科學委員會補助專題研究計畫 5 成 果 報 告
□期中進度報告
數位實驗課程對高中生學習科學的影響
計畫類別：5個別型計畫 □ 整合型計畫 計畫編號：NSC 95－2511－S－011－001－MY3
執行期間： 95 年 8 月 1 日至 99 年 1 月 31 日
計畫主持人：陳素芬 共同主持人：陳秀玲
計畫參與人員：李偉豪、蘇暐珍、麗山高中吳明德老師、旗美高中鄭瑋 凌老師、和平高中王富民老師
成果報告類型(依經費核定清單規定繳交)：□精簡報告 5完整報告
本成果報告包括以下應繳交之附件：
□赴國外出差或研習心得報告一份
□赴大陸地區出差或研習心得報告一份
5出席國際學術會議心得報告及發表之論文各一份
□國際合作研究計畫國外研究報告書一份
處理方式：除產學合作研究計畫、提升產業技術及人才培育研究計畫、
列管計畫及下列情形者外，得立即公開查詢
□涉及專利或其他智慧財產權，□一年□二年後可公開查詢
執行單位：國立台灣科技大學
中 華 民 國 99 年 3 月 31 日
計畫名稱：數位實驗課程對高中生學習科學的影響 計畫編號：NSC 95－2511－S－011－001－MY3 執行期間：95年08月01日 至 99年1月31日
執行單位：國立台灣科技大學 技術與職業教育研究所 主 持 人：陳素芬
共同主持人：陳秀玲
參與人員：李偉豪、蘇暐珍、楊竣宇、麗山高中吳明德老師、旗美高中鄭瑋凌老師、和平 高中王富民老師
中文摘要
本研究的目的是探討 Microcomputer based laboratories(MBL)對高中生在概念學習、科 學態度和科學本質觀等方面的影響。研究的第一階段是設計實驗課程與開發研究工具，針 對三個高二實驗，包括 Newton’s Second Law, Specific Heat Capacity of Metal, and Boyle’s Law，分別發展並測試實驗手冊，實驗器材與教師手冊。實驗手冊經高中教師和二組學生 的修改、測試，文字已相當清楚，然後在麗山高中正式測試。教師手冊則是於教師研習中 測試。研究工具方面，態度量表 30 題含對科學的享受、對實驗的享受，以及對實驗的自我 效能，部分翻譯自 Attitudes Towards Science Inventory (Gogolin & Swartz, 1992)、Science Self-efficacy (Ketelhut, 2004)，以及鄭森榮（2005）「對科學的態度量表」。學生科學本質觀 量表 44 題是由研究者以前針對大學生所設計的量表 Views on Science and Education 改進而 成，取其科學本質分量表。兩量表在旗美高中與和平高中測試(n=416)，Cronbach’s α分別 為.93、.73。教室觀察則使用 Reformed Teacher Observation Protocol 加上部分有關科學本質 的題項。三單元的概念測驗卷的評分者間信度皆>.86。第二階段是比較 MBL 實驗與傳統實 驗的學習成效，以及推廣課程。研究對象為和平高中的兩班高二物理，為常態、男女混合 編班，各有 41 人。一班進行三個 MBL，另一班進行傳統牛頓實驗和 MBL 波以爾實驗。兩 班皆有概念測驗卷與問卷的前後測。此外，在每一次實驗課程中，兩班學生均有錄影，並 且從 MBL 的班級中選一組學生同步錄攝其電腦螢幕影像與小組互動情形，以及做實驗後的 訪談。兩班學生完成實驗後，概念皆有顯著進步，但整體態度與科學本質觀並無顯著不同。
MBL 較省時且與傳統實驗一樣有效地促進概念學習，對實驗的享受與自我效能會隨時間下 滑，但 MBL 似可減緩前者的下滑速度，以及提升實驗自我效能。另外，實驗回饋單、實驗 後的訪談資料以及實驗課程的錄影結果顯示，MBL 能提高學生參與度，引導學生思考更多 影響實驗的因素、做更完整的資料詮釋，並與組員、老師相互討論，進而達到互動與探究 式的學習。最後，於和平高中科學班和社區高中職的科學班推廣所開發之探究實驗課程並 支援師大、國教院與公館國小之相關課程。
關鍵詞：數位量測、高中實驗、科學態度、科學本質、科學探究
II
Abstract
The purpose of this project was to investigate the effectiveness of micro-computer based laboratories (MBL) for high schools students, specifically in their acquisition of content knowledge, attitudes toward science and views on nature of science. During the first stage of the study, MBLs for 11th graders and research instruments were developed and field tested.
Three laboratories, including Newton’s Second Law, Specific Heat Capacity of Metal, and Boyle’s Law, were involved. Each of them is equipped with handheld devices, a laboratory menu and a teacher menu. The curricular materials were tested on two classes of high school students and consulted with two high school physics teachers. We also have developed the teacher’s manual and conducted a workshop for secondary science teachers. The attitude survey and Nature of Science Scale involved 30 and 44 items and Cronbach’s alphas .93 and .73,
respectively. The inter-rater reliability for the three achievement tests was higher than .86.
During the second stage, a comparison between conventional laboratories and MBLs was
conducted with two 11th grade classes. Each class had 41 students. The participants filled the attitude survey, nature of science survey, and achievement tests prior and after the experiments.
All instructions were video-taped. 5-6 students were randomly selected for interviews after each experiment. The results revealed that MBLs are as effective as conventional laboratories
regarding conceptual learning. MBLs may help to maintain students’ enjoyment of laboratories and increase their self-efficacy toward laboratory. The qualitative data showed that MBLs promote students participation and interaction, thinking of alternative experiments, and
interpretation of data. Finally, the MBLs had been adopted by various science classes for high school students, elementary students and pre- and in-service science teachers.
Keywords: handheld device, high school laboratory, attitudes toward science, nature of science, and scientific inquiry
報告內容 前言
本計畫發展以探針（probeware）為主的探究實驗，使學生在正式課程中，意即高中課 綱所載的實驗課程中，體驗科學探究歷程。依照新訂高中課程綱要，配合文獻所建議適合 MBL 的實驗類別，本計畫以高二的三個實驗為主，分別是牛頓第二運動定律、比熱實驗，
以及波以耳定律。設計出三個實驗教案，實驗手冊著重在引導學生思考實驗方法和改進實 驗設計。過程許多需要做判斷的地方都是留給學生決定，讓學生多思考、多嘗試，期能培 養學生的探究能力與邏輯推理能力。這樣的歷程使學生體會學習的樂趣和擁有知識的成就 感，進而使學生具正向之科學態度和認識科學本質。
本研究根據原計畫書完成以下項目：
1. 於麗山高中測試課程
2. 完成教師手冊
3. 測試態度和科學本質量表
4. 發展三份成就測驗卷
5. 於和平高中正式實施課程，收集質量化資料，評量學生之學習成就、對實驗的態
度和科學本質概念。
6. 推廣課程
研究目的
分析現有實驗課程對學生思維的影響，以建構 MBL 之實驗手冊和教師手冊。研究 MBL 課程對學生之概念學習、科學態度，以及科學本質概念的影響。進而提出對未來科學實驗 課程的建議。
文獻探討
除原計畫書中所述之文獻探討，計畫執行當中個人持續擴增文獻探討，以下的文獻分 析將著重於近代數種探究實驗之科學思維，以及研究方法的發展，以提升本研究之嚴謹度 及深度。
The recipe-like laboratory menu has received tremendous criticisms and I shall not recite those criticisms. Rather, the analysis will focus on several approaches that flourish in recent innovative laboratory designs. The analysis helped to construct a framework for the laboratory menu of the current study. Previous studies demonstrated that MBL works more effectively when coupled with an appropriate teaching style. First of all, the effect of MBL can be significantly improved if students are guided by a constructivist approach and a
predict-observe-explain (POE) cycle. Bernhard (2001) evaluated the effectiveness of MBL for
2
the kinematics and dynamics units in three introductory physics courses (class size ranged from 25 to 40) at a Swedish University. He found that students achieved the best learning results when complemented with POE. Similarly, Russell, Lucas, and McRobbie (2004) used MBL with worksheets of POE format to teach 15 11th grade students concepts of thermal physics.
They observed that, being oriented to a constructivist approach, MBL seemed to effectively engage students in deep level cognitive processing and constructing meanings of graphs. Their results highlighted that instructors’ philosophy and teaching styles are as important as the
technological advantages. This study also indicated that MBL settings promoted student interactions and fostered conceptual change.
An alternative approach to POE would be a model building system that allows students to construct models, to simulate the models, and to compare experiential or simulated results.
Hucke and Fischer (2002) investigated 18 physics students’ verbal and non-verbal actions in traditional laboratories, MBL, and MBL with a model building system (MBS). They pointed out that MBS significantly increased the frequency of actions and lab time regulated by abstract cognitions regarding physics. Although students in MBL and MBS settings did not gain more content knowledge as measured by pre and post concept maps than those in the traditional labs, and although they might pay too much attention to the computers to observe some experimental phenomena, they tended to discuss the experimental results immediately after the measurement and find experimental errors quickly.
Finally, online laboratories or simulations are often designed to reduce students’ cognitive loadings (e.g., Korakakis, Pavlatou, Palyvos, & Spyrellis, 2009; Limniou, Papadopoulos, &
Whitehead, 2009; Sweller, van Merrienboer, & Paas, 1998; Winberg & Berg, 2007), which usually leads to strategies such as providing prompts, scaffolding, restricting variables, and visualizing abstract concepts or mathematical structures. Students are observed to have several general problems in a physical laboratory. Hofstein and Lunetta (2004) indicated that they are often occupied by manipulating materials and procedural issues, and do not pay as much attention to elaborating on the underlying theory or construct concepts. Moreover, a high percentage of students manipulate irrelevant variables (van Joolingen & de Jong, 1991) or pay unnecessary attention to trivial matters such as the colors of the wires in simple DC circuits (Finkelstein, Adams, Keller, Kohl, Perkins, Podolefsky et al., 2005). Furthermore, students often focus on getting desirable results, and fail to utilize the complete experimental information or think deeply into the underlying theories (Schauble, Klopfer, & Raghavan, 1991). Designers for digital learning attempt to reduce these distractions/drudgeries by constraining the learners’ interaction with the learning environment or scaffolding an optimal inquiry path for the learners.
While the function of simplifying experimentation is commonly advocated and applied by researchers in the field of technology and science education, some fundamental problems have been overlooked. It is possible that a naïve view of how science is conducted and how scientific knowledge is constructed is conveyed to young learners. A mode of inquiry embedded in most of the above-mentioned strategies is the hypothetical-deductive (HD) method. The HD model is about how hypotheses are confirmed, and depicts the relationship between theory and evidence as deductive (Popper, 1934; Hempel, 1966; Bird, 1998; Dewitt, 2004; Ladyman, 2002). According to HD, observable consequences are deduced from hypotheses, which are then (indirectly)
confirmed or disconfirmed depending on whether the consequences are observed in experiments.
This model is intuitively plausible to some extent, such that many science educators and teachers embrace it unreflectively. For example, in advocating this model, Lawson (2002) analyzed Galileo’s discovery of Jupiter’s moons and several other important scientific discoveries to
illustrate that “many, if not all, scientific discoveries are hypothetico-deductive in nature” (p.21).
He believed that the HD model is the best account of scientific discoveries and is well-supported by models in cognitive science and neurology. Consequently, he suggested that the HD method be taught as the one and only scientific method to secondary school and college students.
HD encounters the Duhem Problem and has been severely criticized in the philosophy of science. Duhem (1906) points out that a hypothesis is able to deduce observable consequences only when it is conjoined with a set of auxiliary hypotheses. These auxiliary hypotheses and background knowledge involve premises, initial conditions of the system, the reliability of instruments, and so on. According to Duhem (1906), “an experiment in physics can never condemn an isolated hypothesis but only a whole theoretical group” (p. 183). The hypothesis to be tested and the auxiliary hypotheses should always be considered as a whole. When
disconfirming evidence occurs, it is always possible to reject or modify an auxiliary hypothesis rather than the main hypothesis. The evaluation of whether the main hypothesis should be abandoned or preserved is not as straightforward as specified by HD. The lesson from the Duhem problem is that hypothesis testing is a holistic matter. By unreflectively accepting the HD model, the existing teaching approaches are in danger of oversimplifying the relationship between hypotheses and evidence.
Science learning would be much more successful if students were taught to consider hypotheses and evidence holistically. In reality, hypotheses may be coincidentally
confirmed/disconfirmed due to the influences of auxiliary hypotheses. Students would learn that, whether or not the experimental results match the textbook or established theories, the roles played by auxiliary hypotheses and conditions should not be neglected, and ought to be examined
4
carefully. In short, holism should be taken into consideration in all cases.
The right approach for laboratories, I suggest, would be to help students understand each experiment holistically, i.e. to evaluate not only the main hypothesis, but also pay heed to auxiliary hypotheses and apparatus. Students would thus have a better chance of learning the right interpretation of the experimental results. Moreover, the holistic approach will enhance students’ ability to solve problems encountered in everyday life, which are often complicated and affected by many factors.
根據此文獻探討，已撰文投稿於 Computers & Education，審查後正修改中。科學態度 的文獻，增加如下：
According to Fishbein and Ajzen (1975) and Zacharia (2003), beliefs affect attitudes, which affect intentions and behaviours. Attitudes are defined as “a mental concept that depicts
factorable or unfavourable feelings toward an object” (Zacharia, 2003, p.793). Attitudes are affected by beliefs and have direct influence on intentions and behaviours. Moreover, attitudes are tenacious and learned. Views on nature of science may significantly influence students’
attitudes (Freedman, 1997). Freedman also points out that hands-on, activity-based laboratory instruction enhances students’ attitudes toward science, which has positive effects on students’
achievement.
Furthermore, gender difference is observed at all levels (6th-10th grades), except the advanced students (Simpson & Oliver, 1985). Boys at all levels have more positive attitudes than girls. Students, except advanced students, become significantly less positive from
beginning to middle of the school year. Their attitudes at the end of the year were slightly lower than the middle of the year, but were not significantly different. Furthermore, students’ attitudes drop as their grades increase. Attitudes drop sharply from 7th grade life science to 8th grade earth science. Science courses for adolescent students fail to draw students to take more science courses.
Osborne, Simon, and Collins (2003) concluded several major factors that influence students’
attitudes toward science. (1) Gender. Negative attitudes are due to lack of understanding of science, which are due to lack of experience in science. Girls would gradually perceive themselves to be better at non-science subjects. More recent studies found that girls are more confident of their ability in science, but choose not to pursue science. Vocational choices to science careers are not appealing to girls. Many science teachers and their students perceive the instrumental value of science, but do not notice the intrinsic cultural value of science (Tobias, 1990). (2) Social economic status (SES) and parental involvement. The influence of SES and parental support on children’s attitudes toward science remains unclear. Researchers have conflicting results. Adolescents’ attitudes are affected by their peers. (3) Classroom environment. Classroom environment is positively correlated with attitudes. “The most positive attitudes are associated with a high level of involvement, very high level of personal support, strong positive relationships with classmates, and the use of a variety of teaching
strategies and unusual learning activities (p. 1066).” The quality of science teaching determines
students’ attitudes toward science. Positive learning environment includes: clear learning goals, make the goals known to students, preview and review the content, help students to contextualize the content in personal experience and link the content to other learning experiences and goals, willing to engage students in deciding learning goals, supportive social context that students are accepted, cared for, and valued, offer a choice from several learning styles and ways of
engagement, and consider students’ circumstances and modify learning tasks accordingly.
Competent and enthusiastic science teachers and a high proportion of students taking science courses are within a self-perpetuating cycle. The effect of curriculum variables is not as clear and significant as that of teaching variables. (4) Stereotype. Science is perceived as a difficult subject that only bright students take science courses. Teachers should emphasize more on the certain loss of not studying science, rather than the positive aspects of studying science. Overall, attitudes are moderately correlated with achievement. The correlation is higher for high and low ability girls. Girls are more motivated to achieve than boys, even though boys have more
positive attitudes.
Five methods are commonly used for measurement of attitudes towards science:
1. Relative popularity of science courses. Let the subjects rank their liking of school subjects (Whitfield, 1980; Ormerod, 1971). It is a relative scale. It cannot be used for measures of attitude change.
2. Q-sort technique. Fifty cards containing positive and negative adjectives are given to the students. Students sort the cards to describe their feelings about science (Freedman, 1997;
Humphreys, 1975).
3. Focus groups: In-depth study of a group of students’ beliefs and attitudes towards science.
4. Liker-scale items. Scientific attitude inventory (Moore & Sutman, 1970) lacks reliability (Munby, 1997). Attitudes Towards Science Inventory (ATSI) (Gogolin & Swartz, 1992) is better. ATSI has 48 items, divided into six subscales: perception of the science teacher, anxiety towards science, value of science in society, self-concept in science, enjoyment of science, and motivation in science. ATSI focuses on science, not school science. Attitude Toward Science Scale (ATSS) (Simpson & Oliver, 1985) has been used for many studies. It has 7 items, reduced from a pool of 30 items. The Cronbach’s alpha is .94. What comes with ATSS is the Achievement Motivation Scale. It has only four items and the alpha is .88.
5. Interviews.
For this project, Likert-scales and interviews were used.
對於研究方法部分，曾就因素分析重新檢視文獻，務使量表嚴謹可用。近年的文獻顯 示，研究者最常用的主成分分析、萃取特徵值大於一之因素，以及正交旋轉法並無依據
（Preacher & MacCallum, 2003; Lorenzo-Seva & Ferrando, 2006; Hayton, Allen, & Scarpello, 2004)。依方法學最近的研究結果，進行因素分析時，宜以平行分析法（parallel analysis method）決定應萃取的因素量。然後以斜交轉軸檢視每一題項在各個因素上的負荷量。本 研究的量表即是以上述建議進行分析。
研究方法
6
一、 MBLs 課程發展：三份實驗手冊經高中教師和二組學生的修改、測試，文字已相當
清楚，然後在麗山高中正式測試。我們特別探討學生對數位量測實驗融入物理課程的 接受度、學生體驗數位量測實驗的感受及數位量測實驗對學生的物理學習成效。參與 測試的是麗山高中兩個升高三的班級，各 30 與 27 人。該校採常態、男女學生混合編 班。學生在實驗課上完後填寫一份實驗回饋單。每次實驗後，選一組學生訪談。根據 觀察、學生的回饋、晤談與實驗記錄進一步修改實驗手冊。
二、 教師手冊：手冊的功能在於使教師體驗探究學習，並思考（1）實驗的意義，（2）
教師應如何引導學生以達到實驗的目的，（3）如何使探究學習發生。目的是藉教師手 冊做教師專業成長，使教師先思考相關問題，促使其從自身的學習經歷反思學習的本 質，然後才導入研究者的想法，MBL 設計理念、解答和 Q＆A 都是放在手冊的後半部。
三、 態度和科學本質量表：本研究的「對科學的態度量表」一開始係依據 Gogolin 與
Swartz(1992) 所 發 表 之 「 對 科 學 的 態 度 量 表 (Attitudes Toward Science Inventory
（ATSI）)」。原問卷將對科學的態度區分為「對科學教師的知覺」、「對科學的焦慮」、
「科學在社會的價值」、「在科學的自我概念」、「在科學的享受」、「科學的動機」六個 面向，每個面向 8 題，共 48 題。研究者依據需求挑選其中三個面向 24 題翻譯修改後 進行預試。先以台北和平高中及高雄旗美高中自然組學生為樣本進行量表的預試。共 發出問卷 429 份，回收問卷 429 份，扣除亂填者及漏填嚴重之廢卷共 19 份，有效問卷 為 410 份，佔總問卷之 95.57%。態度問卷因為是翻譯自 ATSI，故先以 SEM 作驗證性 因素分析，唯本研究資料並不符合 ATSI 原著的 model。最後使用前述之探索性因素分 析方法分析潛在構面。依據因素分析結果及需求刪除 10 題（第 2、5、7、8、11、14、
16、17、20、21 題），其他共保留 14 題。
問卷後續發展欲涵蓋「對科學的享受」、「對科學的動機」、「對科學的自我概念」、「對
科學實驗的自我效能」，除了保留 ATSI 14 題外，依研究需求增加題目，參考鄭森榮
（2005）「對科學的態度量表」共新增 8 題、Ketelhut (2004) 「對科學實驗自我效能量
表」共新增 10 題，再次針對新編製的量表完成測試及信度分析。新發展量表以台北市 和平高中 12 個班級高一學生進行第二次測試。共發出 484 份問卷，回收 484 份，刪除 無效問卷及亂填後剩下 478 份，佔總問卷 98.76%。本量表 KMO 值為.932，Bartlett’s 球形檢定值為 8122.933 (p < .01)，適合進行因素分析，由 parallel analysis method，以 SPSS 模擬 50 data sets，比較實際數據和模擬數據之 eigenvalues，仍然是萃取 3 個因 素。。再以「最大概似法」( maximum-likelihood method ) 萃取方式，運用斜交轉軸之 直接斜交法進行分析。刪除第 9、22 題、剩餘題數為 30 題，每題因數負荷量均有達到.5 以上。
科學本質量表是根據研究者先前發展 Views on Science and Education 的模式研 發，共七個因子，44 題。這同態度量表於 97 年 1 月在和平高中和旗美高中測試，分
別有 245 及 184 位高二或高三學生完成問卷。後於和平高中 12 個高一班級共 484 學生 進行第二次測試。態度量表之 Cronbach’s alphas 均有.90 以上，而科學本質量表均有 大於.70
四、 三份成就測驗卷：由在職教師協助發展牛頓第二運動定律、波以耳定律和比熱等
三份測驗卷，以及其評分規準。試卷主要試測是學生的概念、計算能力和詮釋圖表的 能力。由三位物理教師、研究者，及研究助理五人針對一班學生的答題進行評分，評 分者間的信度分別為.87、.86、.87。問答題和計算題的給分是依其正確程度和可行性 給予 0-4 分。
五、 於和平高中正式實施課程，並評量 MBL 之成效：參與者為和平高中兩班高二學
生，皆為男女混和之常態班級，各 41 名。第一班實施三個 MBL，配合教師課程，於 對應的單元上完後進行實驗，從牛頓實驗到波以耳實驗歷時一年。第二個班進行牛頓 傳統實驗和波以耳 MBL。兩班皆有概念測驗卷與問卷的前後測。此外，在每一次實驗 課程中，兩班學生均有錄影，並且從 MBL 的班級中選一組學生同步錄攝其電腦螢幕影 像與小組互動情形，以及做實驗後的訪談。
結果與討論
三個實驗在麗山高中測試後，學生有相當正面的回應。由 25 份實驗回饋單及與學生訪 談的統計結果，發現有近八成的學生覺得這樣的實驗經驗與以往的實驗經驗是截然不同 的，有七成六的學生覺得比以往的實驗投入，也有近九成的學生喜歡這樣的實驗，所以大 部分的學生對於數位量測實驗是抱持新奇且正面的態度。然而，當問及如何在實驗與上課 中選擇時，選擇想上課的學生有五成；選擇做實驗的僅有三成；二者都選的有近兩成。訪 談中同學表示，這與他們即將面臨升大學的升學壓力有關。
有八成的學生覺得這個實驗有讓他們更清楚相關的物理概念。例如：F=ma、瞭解作用 力、質量、加速度間的關係、PV=常數、PV=nRT；近兩成則瞭解到小組分工合作的重要性；
一成多學會了儀器的操作；另外一成多學會分析實驗誤差來源並改進實驗；其他還有覺得 變因的控制不易、要對數據存有質疑及有人覺得更能體驗實驗樂趣…等。研究者依觀察、
問卷和訪談的結果，進一步修改實驗手冊。
由和平高中實施課程所收集之三個單元測驗卷前後測，顯示無論是 MBL 或傳統實驗，
學生的概念測驗均有顯著進步：
甲班之概念測驗前後分數及成對 t test
前測 M (SD) 後測 M (SD) 成對 t p
牛頓第二運動
定律 MBL 22.55 (4.88) 26.03 (6.28) -4.25 .000
8
金屬比熱 MBL
29.33(9.26) 34.48(8.13) -6.34 .000
波以耳 MBL 20.92 (4.52) 25.53 (3.50) -8.57 .000
乙班之概念測驗前後分數及成對 t test
前測 M (SD) 後測 M (SD) 成對 t p
牛頓第二運動
定律傳統實驗 20.70 (5.83) 23.13 (5.69) -4.62 .000
波以耳 MBL 23.40 (4.75) 26.14 (4.2) -3.637 .001
以前測為共變量的 ANCOVA 顯示，這兩個班的進步分數，無論是牛頓第二運動定律或波 以耳，都沒有顯著不同。
態度部分，對實驗的享受隨時間下降。對乙班而言，於波以耳實驗時使用 MBL，似乎 減緩其下降趨勢。對科學實驗的自我效能，甲班並無顯著改變，但乙班於牛頓的傳統實驗 與波以耳 MBL 之間顯著下降，卻於波以爾 MBL 之後顯著上升(p=.05)。
甲班之科學態度平均值與標準差(n=34)
牛頓前測 牛頓後測 比熱前測 比熱後測 波以爾前測 波以爾後測
M (SD) M (SD) M (SD) M (SD) M (SD) M (SD) 因子一：對科學
的享受與自我 概念
2.52 (.46) 2.52 (.45) 2.41 (.46) 2.53 (.48) 2.50 (.67) 2.49 (.62) 因子二：對科學
實驗的享受 2.84 (.47) 2.78 (.54) 2.72 (.59) 2.64 (.60) 2.71 (.71) 2.62 (.63) 因子三：對科學
實驗的自我效 能
2.12 (.51) 2.16 (.48) 2.14 (.46) 2.24 (.44) 2.25 (.68) 2.23 (.63)
乙班之科學態度平均值與標準差(n=34)
牛頓前測 牛頓後測 波以爾前測 波以爾後測
M (SD) M (SD) M (SD) M (SD)
因子一：對科學的享
受與自我概念 2.45 (.43) 2.47 (.42) 2.49 (.45) 2.54 (.43) 因子二：對科學實驗
的享受 2.80 (.69) 2.69 (.65) 2.59 (.69) 2.59 (.50)
因子三：對科學實驗
的自我效能 2.37 (.49) 2.41 (.44) 2.28 (.50) 2.40 (.39)
科學本質部分，兩班僅在「科學知識的持久性」因子，有顯著進步，t=3.25, 3.14, p<.01。亦即學生更加相信科學知識是不易變革的。但在其他因子並無顯著改變。
以上數據顯示，MBL 與傳統實驗在學生的學習上一樣有效，甚至可能減緩學生因升學 考試的壓力而對實驗逐漸失去享受的情形，與提升學生的實驗自我效能。觀察與晤談資料 較明顯支持 MBL，學生喜歡其方便省時、有趣、簡單、準確、儀器較好，可以體會 2-4 倍 大氣壓力的感覺等。MBL 所配合的手冊雖然較難，但多數學生表示需思考、有挑戰性、有 意義，反而使他們更投入。藉由實驗發現誤差來源並改進實驗、儀器的操作、控制好變因、
更謹慎地做實驗、讓他們感覺較具體、有真實感，有助於加深物理概念。學生也表示 MBL 手冊使小組的討論更實際，組員能分工合作、激發點子、交流想法、分享成就。觀察發現，
在傳統實驗中，學生只需照著實驗步驟做，不需太多思考，多淪為各做各的工作，例如一 人操作、一人記錄數據，有些組員會沒事做，沒有太多討論，導致組員間互動不多。相反 地，此 MBL 因實驗儀器、手冊、模式的不同，透過引導、探究的歷程，讓學生有更多的思 考、討論，一起腦力激盪、團隊合作來共同完成實驗，此與訪談結果吻合。最後，所有做 過 MBL 的學生沒有人比較喜歡傳統實驗。
參考文獻(selected)
鄭森榮（2005）探究式實驗對國小六年級學童科學本質與對科學的態度影響之研究。未出版之 碩士論文，新竹教育大學人力資源教育處教師在職進修應用科學系教學碩士班，新竹市。
Bird, A. (1998). Philosophy of science. Routledge Publisher, London.
Brush, S. G. (2004). Comments on the epistemological shoehorn debate. Science and Education, 13(3), 197-200.
Gogolin, L., & Swartz, F. (1992). A quantitative and qualitative inquiry into the attitudes toward science of nonscience college majors. Journal of Research in Science Teaching, 29,
487-504.
Hayton, J.C., Allen, D.G., & Scarpello, V. (2004). Factor retention decisions in exploratory factor analysis: A tutorial on parallel analysis. Organizational Research Methods, 7, 191-205.
Hofstein, A., & Lunetta, V.N. (2004). The laboratory in science education: Foundations for the twenty-first century. Science Education, 88(1), 28-54.
Hucke, L. & Fischer, H. E. (2002). The linking of theory and practice in traditional and in computer-based university laboratory experiments. In D. Psillos & H. Niedderer (Eds.), Teaching and learning in the science laboratory (pp.205-218). Boston: Kluwer Academic Publishers.
10
Ketelhut, D. J. (2004). Assessing science slef-efficacy in a uirtual environment: A measurement pilot.
(unpublished)
Lawson, A. E. (2002). What does Galileo’s discovery of Jupiter’s moons tell us about the process of scientific discovery? Science and Education, 11(1), 1-24.
Lorenzo-Seva, U. & Ferrando, P.J. (2006). Factor: A computer program to fit the exploratory factor analysis model. Behavior Research Methods, 38, 88-91.
Preacher, K.J. & MacCallum, R.C. (2003). Repairing Tom Swift’s electric factor analysis machine. Understanding Statistics, 2, 13-43.
Russell, D. W., Lucas, K. B., & McRobbie, C. J. (2004). Role of the microcomputer-based laboratory display in supporting the construction of new understandings in thermal physics.
Journal of Research in Science Teaching, 41, 165-185.
Sweller, J., van Merrienboer, J. J. G., & Paas, F. G. W. C. (1998). Cognitive architecture and instructional design. Educational Psychology Review, 10(3), 251–296.
計畫成果自評
在實務面，這個計畫逐漸凝聚有志改進實驗課程的學校和個別教師。主持人在原有的 計畫規劃之外，特書寫教師手冊，且在 96 年 10 月舉辦兩天的教師研習，部分參與的教師 成為長期合作對象。此外，我們試著結合不同團隊，進行推廣，並且依據第一年的實驗設 計模式，為不同學習對象設計實驗課程。推廣活動表列如下：
數位實驗課程推廣活動
日期 地點 對象 探究實驗單元
2007.9.17 國家教育研究院 各縣市自然科輔導員及教師
約 60 人
水沸點量測
2007.10.10 -17
台科大「數位量測實 驗教學研討會」
職前與在職教師約 30 人 牛頓運動定律實驗、金
屬比熱實驗
2007.12.3
國家教育研究院 各縣市自然科輔導員及教師
約 60 人
槓桿、輪軸、定滑輪、
動滑輪實驗 2008.3.5-4.
9
和平高中 高一及八年級 39 人 牛頓運動定律實驗、金
屬比熱實驗、波以耳實 驗
2008.3.8 台師大「語言與科學
學習工作坊」
職前與在職理科教師約 40 人
波以耳定律
2008.3.24 國家教育研究院 各縣市自然科輔導員及教師 波以耳定律
約 50 人 2008.5.26-
6.5
公館國小 國小學生約 20 人 冰鹽實驗、聲音的傳
播、電路實驗、力學（火 箭）實驗
2008.11.5-1 2.17
和平高中 台北市南區各高中、職學生
25 人
2009.10-12 公館國小 國小學生約 20 人 冰鹽實驗、電路實驗、
摩擦力實驗、生理實 驗、水壓實驗 在學術研究面，本計畫下之研究發表與投稿情形如下：
期刊論文：
Chen, S., Hsu, I. C., & Wu, C-M. (2009). Evaluation of Undergraduate Curriculum Reform for Interdisciplinary Learning. Teaching in Higher Education. (SSCI)
Chen, S. (in revision). The view of scientific inquiry conveyed by simulation-based virtual laboratories. Computer & Education. (SSCI)
研討會論文：
陳素芬、李偉豪、吳明德（2007）。數位量測與探究融入高中物理實驗。論文發表於「2007
物理教學及示範研討會」，台中。
李偉豪、吳明德、陳素芬（2007）。數位量測實驗對高中生學習物理的影響。論文發表於
「中華民國第二十三屆科學教育學術研討會」，高雄。
Chen, S.-L., & Chen, S. (2008). College Science Instructors' Views and Experiences of Curriculum Reform. Paper presented at Annual Meeting of the National Association for Research in Science Teaching, Baltimore, USA.
Chen, S. (2009). Hypothetico-deductive reasoning in virtual laboratories. Paper accepted by the Tenth International History, Philosophy, and Science Teaching Conference, Notre Dame, USA.
Chen, H. Y., & Chen, S. (2009). Is Tacit Knowing an Uncertainty or an Answer to Science Education? –Michael Polanyi Revisited. Paper accepted by the Tenth International History, Philosophy, and Science Teaching Conference, Notre Dame, USA.
Chen, S. (2009). The view of scientific inquiry conveyed by simulation-based virtual laboratories.
Paper presented at the biannual meeting of European Science Education Research Association, Istanbul, Turkey.
Chen, S., Su, W.-C., & Li, W. (2010). An investigation of the learning effects of simulation-based
12
laboratories and microcomputer-based laboratories. Paper presented at the Sixth International Conference on Science, Mathematics and Technology Education, Hualien, Taiwan.
行政院國家科學委員會補助國內專家學者出席國際學術會議報告
98 年 9 月 10 日 報告人姓名
陳素芬 服務機構
及職稱
國立台灣科技大學副教授
時間 會議 地點
98 年 8 月 31 日起至 98 年 9 月 4 日止
Istanbul, Turkey
本會核定 補助文號
NSC 95-2511-S-011-001-MY3
會議 名稱
(中文) 科學教育學會
(英文) European Science Education Research Association 發表
論文 題目
(中文)線上模擬實驗所傳達之科學探究觀
(英文) The view of scientific inquiry conveyed by simulation-based virtual laboratories
報告內容應包括下列各項：
一、參加會議經過
八月三十一日至九月四日五天的會議中，本人發表論文並參加知識論、科學本質、
Science Teaching and Learning 和 Educational Technology 等主題的 sessions。這 些主題均與我目前所從事的國科會計畫相關。
二、與會心得
此文章是本人參與的國科會計畫的部分成果，著重在探討既有數位實驗的問題。透過深 度省思而起發新的實驗設計概念。在會議中，透過口頭報告和與會的學者互動，發表後 聽眾討論熱烈，所得到的回饋對後續的實驗課程發展輿論文發表有莫大幫助，也激發我 思考尚待研究的議題。相關的研究結果已投稿 Computers & Education。
三、考察參觀活動(無是項活動者省略)
無
四、建議
這次活動使個人認識近年科教在土耳其的發展，受到極大的激勵，更敦促本人的研究活 動。感謝國科會的支持，使國內科教人士與國際科教界有此交流活動，藉此提高我們的 研究水準。個人也是藉此次會議，持續與國外的合作方案，希望能在研究上更上一層。
五、攜回資料名稱及內容
附件三
14
會議中所發表的論文摘要 數十篇論文、proceedings
六、其他
附件為發表之論文之一
The view of scientific inquiry conveyed by simulation-based virtual laboratories Sufen Chen, Graduate School of Technological and Vocational Education/ Education Center, National Taiwan University of Science and Technology. sufchen@mail.ntust.edu.tw.
Abstract. With an increasing number of studies evincing the effectiveness of simulation-based virtual laboratories (VLs), researchers have discussed replacing traditional laboratories.
However, most VLs convey an oversimplified view of scientific inquiry to students. A survey of the online VLs revealed that hypothetico-deductive (HD) logic prevails in their design. Ever since Duhem and Kuhn, philosophers of science have learned that the relationship between hypotheses and evidence is holistic; however the pedagogical value of this point has not received enough attention in science education. Many science educators and VL designers still
uncritically adhere to the HD method. This article comments on the HD method and how VLs can avoid embracing it.
Keywords: simulation; scientific inquiry; virtual laboratory; scientific reasoning
16
The view of scientific inquiry conveyed by simulation-based virtual laboratories 1. Introduction
The purpose of this article is to examine the reasoning patterns underpinning most simulation-based virtual laboratories (VLs). VL is defined as a simulation environment in which learners interact with virtual apparatus and materials and conduct experiments on a computer. The advance of technology has induced new ways of experimenting, and various techniques and systems have been developed to model authentic laboratories. In a VL, students can manipulate parameters and test hypotheses. A dynamic collaborative system allows
teammates to communicate synchronously and take turns working on a shared real-time experiment (Jara, Candelas, Torres, Dormido, Esquembre, & Reinoso, 2009). The
teacher-student and student-student interactions are fairly intuitive and realistic. Furthermore, VLs may incorporate guidance to reduce cognitive loading and, thus, to facilitate meaningful learning (Holzinger, Kickmeier-Rust, Wassertheurer, & Hessinger, 2009). Guidance can also be utilized to forestall random interaction with simulations and to focus on important parameters and concepts.
VL, being more manageable, safe, cost-efficient, clean, flexible, and rapid than physical experiments (Triona & Klahr, 2003; Zacharias & Constanitinou, 2008), has become a popular alternative to traditional laboratories. Specifically, numerous studies have evinced positive learning effects of virtual environments that support learners to explore, test hypotheses and analyze data as scientists do (Gordon & Pea, 1995; Ramasundaram, Grunwald, Mangeot, Comerford, & Bliss, 2005; McElhaney, 2007; Yang & Heh, 2007; Sun, Lin, & Yu, 2008). A few studies have purposely compared VLs with traditional laboratories and shown the advantages of VLs in concept learning (Triona & Klahr, 2003; Klahr, Triona & Williams, 2007; Zacharias &
Constanitinou, 2008). As a result, an increasing number of researchers are proposing the replacement of physical laboratories by VLs.
Studies in the field have mainly been based on cognitive load theory (e.g., Korakakis, Pavlatou, Palyvos, & Spyrellis, 2009; Limniou, Papadopoulos, & Whitehead, 2009; Sweller, van Merrienboer, & Paas, 1998; Winberg & Berg, 2007), which usually leads to strategies such as providing prompts, scaffolding, restricting variables, and visualizing abstract concepts or mathematical structures. Constructivist beliefs including links to prior knowledge, active learners and meaning construction are often highlighted (e.g., Ramasundaram et al., 2005).
However, the function of laboratories has been simplified to become merely a tool for learning scientific content in most studies. The learning of using instruments, equipment and techniques, and cultivating positive attitudes toward science through VLs have occasionally been recognized, but not necessarily investigated (e.g., Kong, Yeung, & Wu, 2009). Philosophical issues are not noticed.
As an approach to learning and teaching, it is important for VL to lay its foundation on psychological and cognitive needs, such as individual differences and cognitive loadings.
However, for VL to be considered as an alternative to physical laboratories, issues beyond psychology and conceptual learning should be discussed. Specifically, a philosophical lens can help to scrutinize how consistent VL is with the objectives we anticipate to achieve through
science education in general and laboratories in particular. Researchers ought to consider the picture of science that VLs convey to learners and how it influences science education as a whole.
This article examines VLs from a philosophical perspective and focuses on the problems of using the hypothetico-deductive method in VLs. It is hoped that the discussion will contribute to the future design of VLs and induce more discussions of the related philosophical issues.
In short, the implications of VLs would be greatly extended and more productive if the design went beyond psychological concerns, and the program evaluation went beyond the learning of content knowledge. In this article, the hypothetico-deductive model serves as an example to elevate more philosophical discussion in the field. The discussion will evolve a broader spectrum of the design and evaluation of VL systems.
2. Virtual laboratories and the hypothetico-deductive method
Students are observed to have several general problems in a physical laboratory. Hofstein and Lunetta (2004) indicated that they are often occupied by manipulating materials and
procedural issues, and do not pay as much attention to elaborating on the underlying theory or construct concepts. Moreover, a high percentage of students manipulate irrelevant variables (van Joolingen & de Jong, 1991) or pay unnecessary attention to trivial matters such as the colors of the wires in simple DC circuits (Finkelstein, Adams, Keller, Kohl, Perkins, Podolefsky et al., 2005). Furthermore, students often focus on getting desirable results, and fail to utilize the complete experimental information or think deeply into the underlying theories (Schauble, Klopfer, & Raghavan, 1991). VLs can considerably reduce these distractions/drudgeries by constraining the learners’ interaction with the learning environment or scaffolding an optimal inquiry path for the learners. Almost all published VLs for K-12 have been designed in accordance with these principles. VLs physically and mentally simplify laboratory tasks for students.
While the function of simplifying experimentation is commonly advocated and applied by researchers in the field of technology and science education, some fundamental problems have been overlooked. It is possible that a naïve view of how science is conducted and how scientific knowledge is constructed is conveyed to young learners. Moreover, the clean data created by the system often implies a definite solution. Scientific knowledge is reconfigured and reordered in a form that takes no notice of the construction of the knowledge or the possibility for further research. This article focuses on the mode of inquiry embedded in most VLs.
2.1. Method
This research started with browsing the available VLs on Google, Learningscience.org, Egyptian Virtual Labs Portal and the literature. Special attention was paid to VLs for K-12.
Software used by professionals, such as WQMAP (Water Quality Mapping Analysis Package) for water quality modeling, CitiGreen for urban ecology analysis and CommunityViz for community planning, were not reviewed. Simulations that do not allow the manipulation of variables were also excluded. Finally, 233 VLs in physics for K-12 were examined. They were mainly from five websites: Virtual Physics Laboratory (http://www.phy.ntnu.edu.tw/java/indexPopup.html), Java Applets on Physics (http://www.walter-fendt.de/ph14e/index.html), PhET Project
(http://phet.colorado.edu/simulations/index.php?cat=Featured_Sims), MyPhysicsLab
18
(http://www.myphysicslab.com/index.html), and Interactive Physics and Math with Java
(http://www.physics.uoguelph.ca/applets/Intro_physics/kisalev/index.html). In addition, Virtual Labs Electricity
(http://www.wsd1.org/LTCActivities/46%20Freeware/virtual_labs_electricity.htm), Thermo lab (http://www.chemcollective.org/applets/vlab.php) and Concord Consortium
(http://www.concord.org/) were reviewed.
For each VL, whether it simulated ideal conditions was identified. In K-12 science, typical ideal conditions include motions free from friction and any dissipative effects, elastic collisions, non-interacting gas particles, point masses, and perfectly sealed systems. Variables that can be manipulated, experimental error sources and restriction were coded. Moreover, there existed many lesson plans uploaded by users on PhET. The lesson plans for two randomly selected VLs, Projectile Motion and Pendulum, were analyzed to elucidate the logic of teaching.
2.2. Results
The results revealed that 80% of the VLs are ideal cases. In other words, motions are friction- and air resistance-free. Collisions are elastic. Wave is presented without any noise and propagates without loosing energy. The other 20% simulations allow learners to manipulate friction coefficients, drag force, elasticity, fluid viscosity or tension of strings to observe their effects. Only two VLs in PhET involve some restrictions that are not salient to learners. For 99% of the VLs, students need not be concerned with error sources nor reflect on experimental designs.
Although learners can set up experimental conditions in those 20% simulations, ideal experimental conditions are always the default. Furthermore, as shown by the lesson plans uploaded by teachers on PhET (http://phet.colorado.edu/teacher_ideas/browse.php), teachers normally ignore friction, air resistance and restriction of experiments. For example, among the 25 lesson plans available for projectile motion, 16 literally ask students to turn off the air
resistance when testing the effect of a variable. The other lesson plans either treat air resistance as a variable or do not comment on it. Two of the lesson plans direct students to look at air resistance as an error source:
In my course, the students design their own labs and there is a strong focus on identifying and minimizing error sources. … I also used some things that are more affected by air resistance like paper, balloon, and cotton ball (Loeblein, 2008c).
Examine under what circumstance the air resistance can be ignored (Rott, 2008).
However, no further description is provided. Moreover, for the pendulum simulation, the equation T=2π√(L/g) works only for small angles or amplitudes. Nevertheless, the available lesson plans, except those written by the design team, overlook the angle restriction.
These idealized simulations denote a definite relationship between variables. Drawing a conclusion from their experimental results is fairly straightforward. The laboratories are therefore organized in a logical step-by-step process (O’Byme, 2008). Many instructions are configured in a prediction/hypothesis-experimentation-conclusion layout (e.g., Ishimoto, 2008;
Kennedy, 2008; Workshop, 2009a, 2009b; Wright, 2007), in which the conclusion is often recognized as confirm/prove or disprove a prediction/claim (McCurdy, 2008a, 2008b; Wright,
2007). Students are not guided to make estimations, inferences, explanations, modifications of experiments or examinations of anomalous data, although these tasks are constantly dealt with in an authentic experiment.
With the idealized simulations, a mode of logic for learning and teaching has been molded.
This logic is unambiguously specified by Limniou et al. (2009) for a pH simulator centered on acid-base titrations:
Special care was taken as for students… to confirm or reject their initial hypothesis…. By using the simulator students collected the experimental data easily and quickly in order to confirm or reject initial hypothesis (p.48).
In the philosophy of science, this is identified as the hypothetical-deductive model.
2.3. Hypothetical-deductive (HD) model
The HD model is about how hypotheses are confirmed, and depicts the relationship between theory and evidence as deductive (Popper, 1934; Hempel, 1966; Bird, 1998; Dewitt, 2004;
Ladyman, 2002). According to HD, observable consequences are deduced from hypotheses, which are then (indirectly) confirmed or disconfirmed depending on whether the consequences are observed in experiments. This model is intuitively plausible to some extent, such that many science educators and teachers embrace it unreflectively. For example, in advocating this model, Lawson (2002) analyzed Galileo’s discovery of Jupiter’s moons and several other important scientific discoveries to illustrate that “many, if not all, scientific discoveries are
hypothetico-deductive in nature” (p.21). He believed that the HD model is the best account of scientific discoveries and is well-supported by models in cognitive science and neurology.
Consequently, he suggested that the HD method be taught as the one and only scientific method to secondary school and college students.
Lawson’s analysis has been questioned by historians of science, including Allchin (2003) and Brush (2004), for overlooking the context of generating hypotheses. Allchin ٛriticized Lawson for shoehorning Galileo’s case into the preconceived philosophical stance, HD, which might thus convey an inauthentic nature of science to students. However, something important has been left out of his debate. HD is not about generation of hypotheses, but mainly about justification of hypotheses (Dewitt, 2004). The following discussion will show that HD is problematic in the context of justifying hypotheses. Today, HD prevails in simulation-based VLs. There is a need to critically evaluate this method so that science educators and
VL-designers do not uncritically adhere to it.
3. Problems of HD as a method for testing hypotheses
HD encounters the Duhem Problem and has been severely criticized in the philosophy of science. Any defenders of HD have to take this criticism into account. In the following, I focus on some philosophical and pedagogical problems of HD as a method for testing hypotheses, and argue that a holistic approach should be taken to consider the relationship between
hypotheses and evidence.
Duhem (1906) points out that a hypothesis is able to deduce observable consequences only when it is conjoined with a set of auxiliary hypotheses. These auxiliary hypotheses and background knowledge involve premises, initial conditions of the system, the reliability of
20
instruments, and so on. According to Duhem (1906), “an experiment in physics can never condemn an isolated hypothesis but only a whole theoretical group” (p. 183). The hypothesis to be tested and the auxiliary hypotheses should always be considered as a whole. When
disconfirming evidence occurs, it is always possible to reject or modify an auxiliary hypothesis rather than the main hypothesis. The evaluation of whether the main hypothesis should be abandoned or preserved is not as straightforward as specified by HD. The lesson from the Duhem problem is that hypothesis testing is a holistic matter. By unreflectively accepting the HD model, the existing VLs are in danger of oversimplifying the relationship between hypotheses and evidence.
Usually many details of the auxiliary hypotheses are left out in the practice of science.
When evidence conflicts with prediction, scientists seldom reject or modify the main hypothesis immediately. Rather, they tend to check the auxiliary conditions first to see if they can find problems there. In science classrooms, this is particularly true when experimental results are at odds with established theories. Instructors usually help students to detect errors in experimental conditions and processes. For example, in the case of simple pendulum motion, students might hypothesize that a change in the weight hanging on the end of a string causes a change in the period of the pendulum. Although this hypothesis contradicts what is taught in textbooks, it is nevertheless very likely to be observed by students in their experiments. Should instructors allow students to accept the false hypothesis? Obviously not. Students should be guided to consider other possible explanations. For example, the change in period may in fact be caused by a change in the centre of the mass, angular momentum, friction, and so on. More
experiments are needed before accepting the hypothesis. This holistic approach is valuable because it enables students to avoid the mistake of oversimplification mentioned above.
4. Discussion
Concepts in science, especially in physics, can generally be represented by mathematical models. Most simulations are accurate as they employ mathematical equations to present the relationships between variables (Bryan, 2006; http://modellus.fct.unl.pt/). One can imagine that such a perfect match of scientific laws and data would certainly enhance students’ learning of those laws, compared with ordinary noisy experiments. Yet it should be noted that scientists used the same or less accurate apparatus as our schools traditionally use to develop those scientific laws. The training is not just about learning scientific concepts or some
domain-general skills, but also about how knowledge is constructed and how a problem is holistically inspected. By replacing laboratories with simulations, we may direct students to a naïve thinking path that follows oversimplified logic or HD reasoning.
Science learning would be much more successful if students were taught to consider hypotheses and evidence holistically. In reality, hypotheses may be coincidentally
confirmed/disconfirmed due to the influences of auxiliary hypotheses. Students would learn that, whether or not the experimental results match the textbook or established theories, the roles played by auxiliary hypotheses and conditions should not be neglected, and ought to be examined carefully. In short, holism should be taken into consideration in all cases.
However, the perfect match between data and theory in VLs does not offer chances for such
