• 作者： 林財庫
• 作者服務機構： 高雄師範大學物理系所
• 中文摘要： 本文結合了物質微粒模式之「預試晤談分析、內容分析、認知分析和歷史發展分析」等的研究結果發展了一份（兩段式）診斷中小學生學習「物質微粒模式」之相關概念時所既有或新生之認知樣式（patterns）和層次的雙層試題和半結構性測後晤談問卷。這份診斷工具包含15 個概念，分為四個向度（題組），是經由分析高雄市兩所高中、兩所國中、兩所國小，共九個年級（小四到高三）2467 位學生的半開放式紙筆測驗（題幹選項+ 理由陳述）、8 位預試晤談對象及128 位受測學生之測後晤談資料的結果，分析與歸納而得。本診斷工具的發展在形式上仿照Treagust 的步驟，在內容上則遵循人本建構論者的建議，在建立題庫和題組的過程中反覆參照前人的研究及「物質微粒模式」之認識論分析（包括內容分析、歷史發展分析和認知分析等）的結果。在結合這三種分析的基礎上建立16 種具有高低（或先後）次序的「範疇結構模式」，以此作為物質微粒理論之科學認知模式的理論原型（propotypes），來編碼、分類和排序受測者筆試測驗和測後晤談的「質性資料」，並嘗試提出可以把這個過程一般化（generalized）的方法論原則。本診斷工具以「項目分析」來評估筆試測驗之每個試題的適切性（每一題都有鑑別度）；以「因素分析」和「理論修正」來建立各題組的構念效度；以內容分析和歷史發展分析做為筆試及測後晤談之內容效度的參考；以受測者基本學力測驗或數理（自然、數學或理化等）成績作為效標來建立筆試測驗的效標關連效度；以受測者學力測驗各科成績與本診斷工具之總分的高中低分組差異的變異數分析來建立另一種理論的構念效度（理論預測本診斷工具的總分與邏輯化程度較高的自然科及數學科成績有正向顯著相關，與邏輯化程度較差的社會科與語文科成績沒有顯著相關，預測結果與理論一致）。本診斷工具筆試測驗之整體（小四到高三）的α 信度，理由陳述部分平均為0.80；筆試及測後晤談之理由陳述部分的再測信度平均為0.75。
• 英文摘要： A two step diagnostic instrument, which includes a two-tier paper pencil test and post-test interviews, have been developed to explore Taiwanese Grades 4-12 students' alternative conceptions in learning the particle theory of matter. This instrument includes 15 concepts (table 2), 33 items (Appendix one), and has been divided into four factors (groups). Each item of the written test consists of a content question having four choices, followed by a space for the pupil to complete the reason why the particular choice was selected. The instrument was administered to 2467 pupils from grades 4-12 in the city of Kaohsiung. After the written test, 128 students of the participants, across the ability range in math and physics with the ratio 1:2:1, were interviewed individually and in depth two months later. Four pairs and 16 sets of combinations(2/sup 4/ = 16) of characteristic categories of particle theory of matter (Table 1) have been used to code, classify and sequence the quantitative data collected from the written test and post test interviews. The methodological meaning of this new coding method has been discussed. Evidence in support of the validity of the instrument constructs was obtained through a validation study, which indicates that this instrument has a satisfactory content, factor, construct and criteria-related validity for the purpose of this study. The Cronbach alpha reliability coefficient (r= 0.80 in average) and test-retest reliability coefficient (r = 0.75 in average) suggested that this instrument has acceptable internal consistency.
• 中文關鍵字： 物質微粒模式、迷思概念、認知模式、模式識別、機械典範
• 英文關鍵字： Alternative Conceptions; Pattern Recognition; Cognitive Model; Particle Model of Matter; Atomism