Validity and Automated Scoring

Abstract

This chapter concerns the automated scoring of answers to constructed-response items as seen through the lens of validity, and shows how to do automated scoring with validity considerations at the forefront. Automated scoring should be designed as part of a construct-driven, integrated system because the interplay among system components is complex and that complexity must be accounted for in scoring-program design and validation. For automated scoring, as for score meaning generally, the validity argument should rest on an integrated base of theory and data that allows a comprehensive analysis of how effectively scores represent the construct of interest and how resistant they are to sources of irrelevant variance. The automated scoring technology is continuously advancing, with a level of sophistication and complexity far greater than that employed in field. Seven assertions were offered about validity and automated essay scoring. One of the most critical of those assertions was that validation should be broad based.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 28 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.