VELC Test の開発の経緯を説明し、実施データから読み取れるテスト（得点）の信頼性や妥当性を、ラッシュモデリングや SEM など、さまざまな角度から検証したものです。
This paper reports on the development and validation of a new English placement test specifically designed for Japanese university students. After discussing problems with EFL placement tests currently used at tertiary institutions in Japan, the paper describes the concepts, specifications, and development process of the new test. It then presents the results of validation attempts based on the response data collected from a sample of 1,800 Japanese EFL learners. The main findings are: (a) the test exhibits a very high Rasch person reliability of .94, with its items well targeted for the intended population; (b) the test consists of items with quite a wide range of difficulty and those difficulties are invariant across sub-samples; (c) the test produced substantially different mean scores for the 18 institutions at which the participants were enrolled; (d) the test exhibits a high criterion-related validity against the TOEIC(R) test, predicting 68% of its total score variance; and (e) the test data show the best fit to a three-factor correlated model. The paper concludes with the interpretation that the evidence gathered illustrates the validity of the new test.
Key words: プレイスメント・テスト, テスト開発，ラッシュ・モデリング，妥当性，構造方程式モデリング (SEM)