VietTESOL International Convention 2026

This study investigated rater severity, inter-rater reliability, and rating processes in the Writing Task 2 assessment within a high-stakes English proficiency examination context. Conducted at an English center in Hanoi, Vietnam, the study employed an explanatory sequential mixed-methods design. In the quantitative phase, seven teachers independently rated ten essays using the official band descriptors. Statistical analyses were conducted to examine scoring consistency, severity patterns, and alignment with official benchmark scores. In the qualitative phase, think-aloud protocols and semi-structured interviews were used to explore how teachers interpreted the band descriptors and justified their scoring decisions. The findings revealed a high degree of inter-rater reliability in the rank-ordering of essays, indicating broad agreement regarding relative essay quality. However, most teachers demonstrated greater severity than the official benchmark scores, although one teacher showed substantially closer alignment with the benchmark than the others. Severity patterns also varied across essays and scoring criteria. Qualitative findings indicated that teachers generally followed similar rating processes, forming holistic impressions before conducting analytic evaluations and relying on the band descriptors as their primary justificatory framework. Nevertheless, differences emerged in the evaluative weight assigned to format conformity, the interdependency applied across scoring criteria, and the benchmarking practices used to calibrate internal standards. The study potentially contributes to understanding teacher rating behavior in high-stakes writing assessment contexts and offers practical implications for assessment literacy development and professional learning among English teachers in Vietnam.