Investigating the quality of a high-stakes EFL writing assessment procedure in the Turkish higher education context

Employing G-theory and rater interviews, the study investigated how a high-stakes writing assessment procedure (i.e., a single-task, single-rater, and holistic scoring procedure) impacted the variability and reliability of its scores within the Turkish higher education context. Thirty-two essays written on two different writing tasks (i.e., narrative and opinion) by 16 EFL students studying at a Turkish state university were scored by 10 instructor raters both holistically and analytically. After the raters completed the scoring procedure, semi-structured individual interviews were held with them to gain insight into their views regarding the quality of the current scoring procedure. The G-theory results showed that the reliability coefficients obtained from the current scoring procedure would not be sufficient to draw sound conclusions. The quantitative results were partly supported by the qualitative data. Important implications were discussed to improve the quality of the current high-stakes EFL writing assessment policy.