國家教育研究院

　　本研究主要以模擬實驗方式探討不同次級量尺分數（subscale score）計算方法於不同情境中，水平等化與垂直等化在平衡不完全區塊設計（balanced incomplete block design, BIB）之估計效果。此外，延續以前作過的研究，本研究新增多向度試題反應理論（multidimensional item response theory, MIRT）來作為次級量尺分數之計算方法，並探討題內多向度與題間多向度兩種多向度測驗。本研究使用模擬資料估計不同次級量尺分數計算方法在等化測驗之測驗分數估計誤差，研究變項中考慮六個因素︰1.施測人數、2.次級量尺個數、3.次級量尺相關程度、4.受試者能力分布、5.定錨試題配置、6.等化估計方法。

　　The purpose of this study is to explore the estimation effect of horizontal equating and vertical equating in BIB design. Additionally, MIRT method is used to estimate subscale scores. Within-item multidimensional test and between-item multidimensional test are investigated in the research. Using simulation data, the study will estimate the accuracy of subscale scores by different methods. In equating testing design, six factors taken into consideration include sample sizes, subscales sizes, correlation between subscales, distribution of abilities, collocation of anchor items, and equating methods.