Benchmark rating procedure, best of both worlds? Comparing procedures to rate text quality in a reliable and valid manner

Abstract

Assessing students’ writing performance is essential to adequately monitor and promote individual writing development, but it is also a challenge. The present research investigates a benchmark rating procedure for assessing texts written by upper-elementary students. In two studies we examined whether a benchmark rating procedure (1) leads to reliable and generalisable scores that converge with holistic and analytic ratings, and (2) can be used for rating texts varying in topic and genre. Results support evidence that benchmark ratings are a valid indicator of text quality as they converge with holistic and analytic scores. They are also associated with less rater variance and less task-specific variance, leading to reliable and generalisable ratings. Moreover, a benchmark scale can be used for rating different tasks with the same reliability, at least when texts are written in the same genre. Taken together, a benchmark rating procedure ensures meaningful and useful information on students’ writing.