Abstract
The main purpose of this study is to examine how valid peer assessment can be in a high school setting in Japan. One hundred fifty seven students and five teachers participated in this study. The observed scores of the peer and teacher assessments were analyzed using Multi-faceted Rasch Measurement. According to the Rasch analysis, 94% of students did not show misfitting behavior, which suggested acceptable validity of peer assessment (Myford & Wolfe, 2003). Peer assessment was also examined at a class level, and this also fitted into the model. In addition, characteristics of misfit raters were investigated in terms of unexpected responses and English proficiency level, but no specific characteristics in common were found. However, some overall tendencies for peer assessment were observed. The average scores of peer assessment were higher than those of teacher assessment, whereas the standard deviations of peer assessment were smaller than those of teacher assessment. These results indicate that peer assessment of Japanese high school students has reasonable validity in a certain context, although student raters tend to be more lenient compared to teacher raters, and they tend to avoid extreme scores.
