「以知識結構為基礎之適性測驗」（knowledge structure based adaptive testing,KSAT）在建立知識結構與節省試題上有相當成效。本研究使用Sternberg (1977)以成份分析發展出的類比推理能力測量模式為基礎自編之圖形類比推理測驗，搭配「以知識結構為基礎之適性測驗」建立圖形類比推理能力的專家知識結構，依此理論建立試題，委請專家學者依此結構協助建立預試題目並進行預試。依據施測結果分析學生的知識結構，進行信、效度分析後，依此結構建立圖形類比推理測驗正式試題。在經過正式施測後，對資料進行分析，並摘述研究結果如下： 一、建立學生在圖形類比推理能力的知識結構。 二、圖形類比推理測驗在預測精準度0.9 與0.95 時分別可節省54.34%與34.78%的施測題目。 三、以知識結構進行命題，可以明確的診斷出學生在圖形類比推理能力的錯誤概念。

In this knowledge-based economy era, (knowledge structure based adaptive testing, KSAT) it is very effective in establishing knowledge structure and economizing the amounts of numbers of the test. The study aims to employ the knowledge structure to illustrate students’ individual abilities and get up to save the test numbers, which based on components-analyzed analogy reasoning figure test and revise it from Sternberg (1977), and collocate “adaptive test based on knowledge structure” to set up the knowledge structure of abilities of analogy reasoning figure test. Invite scholars and experts to help to establish previewed. Analyze student’s knowledge structure on testing result. Establish formal testing questions of analogy reasoning figure based on the structure of analysis of effect and reliability. After testing officially, the data is analyzed and briefed as: 1. Establish the knowledge structure of student’s ability on analogy reasoning figure. 2. 34.78% to 54.34% testing numbers can be saved when the prediction precision of analogy reasoning figure test is 0.9 to 0.95. 3. The fault concept of student’s ability on analogy reasoning figure can be found out definitely when the thesis is based on knowledge structure.