題庫的建構與試題曝光率控制法的使用,已是電腦適性測驗(CAT)裡減緩試題機密憂慮的研究焦點。建構優質的題庫是持續進行的工作,有時,或許需對題庫作些小幅的更動。本研究針對使用SLC試題曝光控制法,探討淘汰單一試題以及不曾被選用施測試題之後,對於題庫存留試題之試題曝光控制參數可能造成的影響。整個研究採用電腦程式模擬電腦適性測驗程序進行。結果指出,試題之間存曝光控制參數已不適合更動過的題庫,重新建立試題曝光控制參數是必要的工作。另外,淘汰不曾被選用施測的試題似乎對於SLC法在控制試題曝光率以及測量精確性上的表現沒有影響。本研究亦針對不曾被選用施測試題的特徵進行檢視,並分別與高、中、低曝光程度的試題進行比較。研究的結果促進了對試題的交互作用影響試題曝光控制參數的了解,並且提供了測驗研究與實務工作者在建構優質題庫時的有用訊息。
Both pool management and exposure control method implementation have been the focus of easing test security concerns in CAT. Maintaining pools of high quality is an ongoing process and sometimes only a small modification may be needed. This study explored how the deletion of a single item and the unused items might alter the exposure control parameters of the remaining items derived by the SLC algorithm. Simulations were carried out for this study. The findings indicated that the original exposure control parameters were no longer appropriate for a modified item pool and the derivation for the exposure control parameters ought to be repeated. Removing the unused items seemed to cause no effects on the performance of the SLC procedure in the observed exposure rates and measurement precision. The characteristics of the unused items were examined and compared with those of the high, moderate and low exposure groups, respectively. Results from this study have offered a better understanding of the effects of item interactions on the values of the exposure control parameters. Also, this study should provide psychometric researchers and test practitioners valuable insights for a better pool construction design.