任何評量工具在使用前都應建立信、效度，故若要找出數學課室學習中具有優點的概念再圖評量，便需大量有關信、效度問題的實徵研究。本文利用Goldsmith, Johnson, & Acton (1991)與Jonassen, beissner, & Yacci (1993)對於知識結構評量步驟的分類，探討在「引出知識」此一步驟覆所採方式對數學課室學習中概念圖評量常用計分方式之信、效度的影響，而引出知識的方式包括評量者所疌供概念數的多寡以及概念構圖中是否需要受評者另外產生概念的技術。本究先針對南部某技術學院專科部四個班共200位學生進行概念構圖的訓練，評量者再將所提供的概念數分成5、10、15、20等四種，每種皆隨機分配給一個班。每班學生皆指數及其運算和指數函數這兩個單元進行概念構圖，並由兩位評分者進行評分。結果顯示評量者所提供概念數的多寡對概念構圖信度的影響，大部不正單純的線性關係，而且不僅概念數的多寡會影響得分的變異成分的大小，所供概念的內容或品質也正重要的影響因素。評量者若是提供足夠多的概念，概念構圖的信度都不差。是否需要受評者另外產生概念對概念構圖信度的影響程度和評量者所提供概念有關，概念數較少時，其影響很大;概念數較多時，影響不多。在效度方面，不論提供概念數為何，關係計分法及結構計分法都有不錯的聚斂效度。提供概念數變多時，概念構圖評量的內容效度才會變佳。而需要受評者另外產生概念的技術，可以改善概念構圖評量的聚斂效度，此時內容效度也較佳。

　　　　　 Because of reliability and validity of any assessment method must be constructed before being used, we need many empirical studies about the reliability and validity of concept map assessment in methematics classroom learning if we want to find an advantaged concept-mapping assessment. In this article we employ the assessment steps about the structure of knowledge in the arguments of Doldsmith, Johnson, & Acton (19910 and Jonassen, Beissner, & Yacci (1993) to probe into the effects of knowledge elicitation method on the reliability and validity of common scoring method of concept-mapping assessment in mathematics classroom learning. And the knowledge elicitation method we proposed including the number of concepts provided by assessor and the technique in which the students are asked to draw beyond the number of provided concepts or not. At first, 200students of 4 classes in a technical institute of the southern Taiwan were drilled in the concept mapping. E assign test various number of concepts, such as 5、10、15、20, provided by the assessor to each class randomly and then all the students carried out concept mapping for the two units, "exponent & exponential operation" and "exponential function". Two raters evaluated all maps respectively. Results indicated that the reliability of concept-mapping assessment is fine if the number of concepts provided by the assessor are sufficiently large but not effected linearly by the number of concepts. The amount of variance component of the map score was affected not only by the number of concepts but also by the content or quality of concepts. The effect of the asked drawing technique on the reliability of concept-mapping assessment depends on the number of concepts provided. It has enough effect to the reliability when the assessor provides less concepts, whereas hasn't when provides more. In the aspect of validity, no matter what the number of concepts is, the convergent validity of relational scoring method and structural scoring method are both fine. The content validity of concept-mapping assessment becomes finer as the number of concepts becomes larger. The technique in which the students are asked to draw beyond the number of provided concepts can improve both convergent validity and content validity in the same time.