本研究之研究目的在於評鑑兩種多重學科評量工具之建構效度。其中一種評量工具係採用選擇題方式,而一種則以問答題方式來評量學生之多重學科能力之高低。假如某學科在前後兩年施測分數之相關高於其與其他不同學分數之相關,該學科則具較佳之建構效度。本研究發現,以選擇題方法之評量工具,在五種不同學科之評量當中,其中兩種學科之評量達到較佳建構效度之標準。相對地,以問答題為方法之評量工具,在六種不同學科之評量中,僅有一學科之評量工具呈現較佳之建構效度。本研究再運用建構模式來評鑑兩種評量工具之區別效度與聚斂效度,結果顯示聚斂效度存在於此兩種評量工具,但沒有照顯之證據支持區別效度。
The primary objective of this study was to examine the construct validity for the two multiple-content testing programs, the multiple-choice Comprehensive Tests of Basic Skills (CTBS/5) together with the performance-based Maryland School Performance Assessment Program (MSPAP), by evaluating the true-score longitudinal associations among multiple-contest scores in one school district. The following criterion was closely examined: the true-score correlation between two time-period measures of the same content area is higher than its longitudinal true-score correlations with other content areas. This criterion was achieved in two (Reading and Mathematics) of five CTBS/S content subtests, as well as one (Language) of six MSPAP content subtests. The structural equation modeling has been conducted on a multitrait-multimethod correlation dataset, where the traits of Reading and Mathematics were assessed by MSPAP and the old version of CTBSI4. Although convergent validity existed in these two measures, there was little evidence to support discriminant validity in both measures.
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。