2009 年 20 巻 p. 161-169
The present study adds another perspective from item response theory in data analysis to Hinkelman & Grose (2004) which showed the feasibility of using moodle for in-school placement testing, and shows the feasibility and the advantages of using moodle for in-school placement test. In addition, the possibility of the development of a moodle-based computer adaptive test (CAT) is discussed. First, a pilot test consisting of three subtests (vocabulary & grammar, listening comprehension with dialogue, and listening comprehension with monologue) was administered to 268 Japanese students in order to calibrate item difficulties. A Rasch model was utilized to eliminate inappropriate items from the test using fit indices. Each subtest was divided into four 10-minute testlets and administrated separately. After trimming the misfitting items, a 68-item tentative placement test was constructed, which was found to be reliable (KR-20=.90). Then, three item banks for each subtest were built for the future moodle-based CAT. Finally, item equating steps for the future, and the logic of the future moodle-based CAT were illustrated based on the previous studies in area.