當教師新接一個班級時，如果能有一個快速有效的方式，知道班級學生能力的排序情形，將可以幫助教師儘快了解學生，並及時給予適當的教學方式或補救教學。試題分析除了可讓教師了解試題的信度與難易度外，更可提供教師評量試題的鑑別性資訊，利用鑑別性高的試題來施測，可以避免出現太難或太簡單之試題，在本研究中將鑑別性資訊加以運用，試圖選取較少量的試題以區分出學生真實的能力。另一方面，教師可以藉由長期的觀察學生的表現、參酌考試的成績及智力測驗成績作為參照成績。本研究是從甲校五年級九個班經三次定期評量選取部份優良試題，再以兩者的Spearman、Kendall等級相關性及Kappa一致性等方法作分析比較。在九個班級的實例研究當中顯示，以參照成績為基準時，其中有七個班級都是部分優良試題成績的兩種等級相關均大於全部試題；Kappa一致性分析也是有類似的結果。根據研究結果，只要施測少量鑑別性高的試題，即可達到區分學生能力的效果，可以縮短施測的時間，也減輕老師與學生的負擔，再將這些試題累積成題庫，更可當作日後教師命題的參考，也可以推廣給更多學生練習，嘉惠更多的莘莘學子。 Item analysis is a good way for teachers to investigate the reliability and difficulty of the taken examinations in order to improve one's own teaching methods as well as enhance the student's learning efficiency. To choose more suitable items for examinations using high-discriminating information is good for students. In this study, we attempt to select parts of items from the taken examinations to evaluate students' true ability. It forms the partial items for scoring. On the other hand, students' reference scores for comparisons are constructed. Based on these, an empirical study was conducted on nine 5th grade classes in an elementary school. Then, we make comparison by means of Spearman's coefficient of rank correlation, Kendall's coefficient of rank correlation and Kappa coefficient of agreement. It shows that the scores of partial items perform well than the scores of original items according to the reference score. So,we can achieve the effect of distinguish students' true ability by adopting few high discriminating examinations.
