Applying the Rasch model in the validation of a test of metalinguistic knowledge

概要

論文の詳細を見る
The purpose of the current study was to examine the construct validity and unidimensionality of a test offirst language metalinguistic knowledge. A ten-item multiple-choice test of Japanese metalinguisticterminology was designed and administered to 100 Japanese university students. The results of a Raschanalysis indicated that the calibrated item estimates were sufficiently accurate and precise—all of the itemsexhibited good fit with the Rasch model. Implications for test revision focused on the possibility of revisingsome items and increasing the number of items to provide fuller construct coverage and higher reliability andseparation estimates.