초록 열기/닫기 버튼
The purposes of this study are 1) to explore essay selection process to find out representative writing examples for a four-point rating scale, and 2) to validate rating scales based on the analysis of the key features of essays selected. In performance assessment, assessors have been confronted with challenging situations due to abstract and ambiguous scale descriptors, and insufficient rating practices. Benchmark essays (sample essays) are an alternative for assessors to improve the rating validity and to maintain their consistency. The rating data were derived from multiple ratings of seven assessors on 38 essays. Using a four-point scale with assessment criteria, assessors first gave scores on holistic score, then analytical scores on content, organization, and language use. The collected data were analyzed using the Many-Faceted Rasch Model (MFRM) and K-mean cluster analysis. The results showed that seven out of eight essays were representative for the rating scale, with one essay showing the unsuitable misfit statistic. Scale descriptors were modified based on the analysis of essays chosen. Implications of the findings were discussed in reference to the modifications of rating scale descriptors, improving assessor reliability, and writing instruction in EFL.
키워드열기/닫기 버튼
EFL writing assessment, selection of benchmark essays, validation of rating scales, many-faceted Rasch Model, cluster analysis
