本研究最主要的目的是為瞭解DIF在小樣本情境下的偵測效果,並檢測其方法在實際資料上之效果。研究分為兩個部分:研究一為模擬研究,研究二利用實證資料進行DIF分析。研究一主要目的在於瞭解不同樣本人數、不同方法等變項在各操弄情境下對於偵測效果的影響,結果顯示若控制在能力有差異的情況下,對於焦點組人數較少(如100人)時,採用MH-2 搭配樣本人數100/1000(焦點組/參照組)的組合具有較佳效能。研究二參考研究一的結果,對於2005年第一次國中基測數學科視障生和普通生的實際作答反應進行DIF分析,結果顯示該次測驗中有兩題產生DIF,文中對於可能原因予以解釋。
The purpose of the research is to explore the efficiency of DIF detection when the sample size of focal group is small. There are two parts in the research; Study 1 is a simulation study and Study 2 is an empirical study. The aim of Study 1 is to examine the efficiency of DIF detection in different sample size ratios of reference and focal group and in different methods. The dependent variables are power and type Ⅰ error. The result of Study 1 suggests that MH-2 combined with 100/1000 reach the DIF detecting power of 0.703 and type 1 error of 0.073 when ability distribution is controlled unequal. Study 2 is an empirical study for detecting the DIF item between visual impaired and general students of the basic competence test for junior high school students in BCTEST. The result of Study 2 suggests that there be two DIF items in 2005 BCTEST, and discussion about the reason causing DIF is also provided.