本研究目的，主要在探討不同測驗難度情境下，相較於傳統固定通過分數，對於援用最大測驗訊息量法於設定精熟標準時，可能受影響的程度。此外，為便於詮釋設定結果，本研究提出換算古典測驗分數法以進行IRT-θ能力值與古典測驗分數間的轉換，並深究於各類型難度測驗中，可能產生的效果。經分析後，研究結果顯示，當發生不同測驗難度時，若援用傳統固定通過分數進行分類時，會因受試者極端偏態之得分分配，致使易發生較差的精熟／未精熟者分類一致性。相較於最大測驗訊息量法，因本身具備能力估計穩定的特性，則能有效減低影響的層面，整體分類百分比一致性仍具一定水平，大致能維持80%以上。而在分數轉換方面，於不同類型測驗下，轉換效果大致可維持於95%分類一致性水準內，若進一步從樣本分配的角度視之，最大測驗訊息量法因具備隨測驗難易，調整通過標準的特性，相互搭配更有助於在極端難度測驗峙，有效減低分數轉換時，可能導致之錯誤分類精熟／未精熟者的人數。最後，綜整本研究結果提出幾項建議，供未來研究者與實務工作者參照運用。

The purpose of this study is to investigate what degree of influence may happen in applying the mastery standard deriving from maximum test information approach to classify mastery/non-mastery in contrast to classical fixed passing score under diverse test difficulty condition. In addition, for the convenience of test result interpretation, we propose the transformed classical test scores approach to convert IRT-θability to clas­sical test score and try to discover the effect of score transformation in different types of difficult tests. After analysis, we find that we may have extreme skewed score distribution in using classical fixed passing score and produce ill classification of mastery/ non-mastery. In contrast, the maximum test information approach may effectively re­duce the degree of influence to maintain an acceptable 80% exact classification performance because of its characteristic of stable examinees' ability calibration. In the aspect of score transformation, the effect may hold in approximately 95% exact classification in different types of difficult tests. Furthermore, from the view of empirical sample distribution, the result indicates that we may effectively lower the number of error classification deriving from score transformation in accompanying using maximum test information approach when extreme test difficulty happens. It is due to the maximum test information approach possesses the characteristic of adjusting mastery standard depending on different test difficulty. Finally, some conclusions and suggestions are proposed for future usage.