Answers are grounded in this paper and cite their sources.
Most standard setting methods can be classified as either test-centered or examinee-centered. Test-centered methods (e.g., the Angoff method, 1971) appear to work well with objective tests but seem less useful with extended-response tests. Examinee-centered methods (e.g., the borderline-groups method and the contrasting-groups method), on the other hand, appear to be particularly appropriate for extended-response tests. In this article, we describe a generalized examinee-centered method for setting multiple cutscores on a test involving both objective and extended-response items. Judges evaluate a representative sample of examinee performances using a rating scale that is defined in terms of performance standards (e.g., proficient, advanced levels). These ratings are linked to examinee's test scores to generate a functional relation between scores and ratings, which is then used to assign a cutscore to each performance level. This approach is potentially more efficient than traditional examinee-centered methods because all ratings are used to define each cutscore. An example involving the setting of multiple cutscores for a state testing program is presented and used to suggest some ways to evaluate the sampling error in the resulting cutscore.
No takes yet. Share an insight, caveat, or question.
Cohen et al. (1999) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: