本研究是以模擬降溫法來設置兩種自動化編製平行測驗程序,其中一種的選題依據係使用古典理論模式的試題難度指標與鑑別度指標,另一種的選題依據是使用IRT模式的試題難度參數與鑑別度參數;每一種模式各組兩個平行測驗,每個測驗之題數為六十。研究結果顯示,兩種模式編製平行測驗的結果都相當好,而且在測驗得分之各項統計值上也非常接近。本研究顯示在設置自動化編製之程序上,使用古典模式的結果與IRT的結果一樣的好,然而使用古典模式能大量減少計算變項與簡化計算量,降低計算無解的可能。模擬降溫法的編製邏輯配合古典模式的使用,是自動化編製平行測驗的另類選擇。
The study implemented two procedures of automated construction of parallel tests based on the method of simulated annealing. The classical item indices were used to select items in one procedure, and the IRT item parameters were used in the other procedure. Each procedure constructed two parallel tests with each parallel test having sixty items. The results of constructing parallel tests were very good for the two procedures, and their test score statistics were very close. The study showed that, in the automated construction procedures, using classical model is as good as using the IRT model; however using classical model can reduce the number of variables and the amount of computation in the procedure to a large extent, and the probability of infeasible solution as well. The implementation of the method of simulated annealing with the application of the classical model has been shown to be a good alternative to constructing parallel tests.
為了持續優化網站功能與使用者體驗,本網站將Cookies分析技術用於網站營運、分析和個人化服務之目的。
若您繼續瀏覽本網站,即表示您同意本網站使用Cookies。