Score Generalizability in Writing Assessment

The book "Score Generalizability in Writing Assessment" by Atta Gebril offers a comprehensive analysis of the reliability of writing assessments in the context of second language evaluation. It highlights the various variables that influence writing scores, such as task prompts, raters, assessment rubrics, genres, topics, and the features being evaluated. The classical method for estimating reliability is deemed inadequate, as it fails to differentiate between the various sources of error. In contrast, generalizability theory provides a well-founded alternative by offering detailed information about the different sources of error and examining their interactions in the assessment process. The book takes an interdisciplinary approach, combining applied linguistics and psychometric research to analyze the relative importance of tasks and raters for writing scores and to propose possible scenarios for maximizing measurement accuracy.