以答題信心度為基礎之線上診斷評量系統

Development of an On-Line Learning-Problem Diagnostic System Using a Confidence-Scoring Mechanism

一般線上評量系統的測驗方式，是由學生回答單選或複選題；由於忽略了學生答題過程的不確定因素，因此在診斷學生的學習困難時，不易作出精確的研判。針對這個問題，已有學者嘗試以答題信心度為基礎，來開發線上評量系統，並獲得不錯的成果。然而，由於目前已提出的評量模式，是由學生主觀認定其信心度，因此在診斷的準確性方面仍有很大的改善空間。在本研究中，我們嘗試加入答題時間參數，來加強其診斷效果，以期更準確地分析學生遭遇的困難。為了驗證這個新評量機制的分析成效，我們以國小學童中文字辨識課程進行實驗；實驗結果證明，本研究提出之「考量時間因素之信心度診斷評量機制」較原有的機制能提供更精確的診斷結果。

In conventional testing systems, students usually take a test by selecting one or more answers for each test item. Nevertheless, the confidence degree of individual student in answering each test item is usually ignored, which might affect the accuracy of diagnosing student learning problems. Researchers have attempted to cope with this problem by developing test systems that allow students to show their degree of confidence regarding the answer to each test item. Such a confidence scoring mechanism has been proved to be effective. However, the existing confidence scoring mechanism is mainly based on the subjective judgments of the students, which may reduce the degree of diagnostic accuracy. In this study, we attempt to develop a new confidence-scoring model by taking the “time spent answering each test item on the Internet” factor into consideration while determining the conference degree for individual students to answer each test item. To evaluate this innovative approach, an experiment has been conducted using a Chinese word- identification course. Experimental results showed that our approach can provide more accurate learning suggestions to students, and hence is more helpful in improving their learning performance than previously proposed approaches.