Latent Growth Curve Analysis of a Trend in IRT English Scores : An Introduction of LGM to Language Assessment Research

概要

論文の詳細を見る
Latent growth curve modeling has become an increasingly accepted method of longitudinal analysis. This study applied an LGM to the longitudinal data of IRT scores from the Ibaraki Prefecture English Test (IPET), one of the prefecture-wide English achievement tests. The scores of approximately 140,000 test takers taking the IPET from 1995 to 2002 became comparable on a common scale by "post-hoc IRT equating" (Saida & Hattori, 2008). The current study focuses on mean scores from schools that participated in the testing every year. The first finding is that the trends in the mean scores fell into three groups: 1995 to 1997, 1998 to 2001, and 2002. This suggests a prefecture-wide decline in some aspects of English language ability following the implementation of the 1989 revision of the Courses of Study. The second finding is that schools with higher mean scores in 1995 showed a slower rate of decline than did schools with lower mean scores in 1995. This indicates that there has been a slightly widening gap in scores across schools over the eight years. This study demonstrates that latent growth curve models (LGMs) are powerful analytic tools for investigations of individual growth and patterns of changing language performance.
社団法人大学英語教育学会の論文
2009-10-20