Automated Scoring of Paper-and-Pencil Figural Responses

Large-scale testing is dominated by the multiple-choice question format. Widespread use of the format is explained, in part, by the ease with which multiple-choice items can be scored automatically. This article examines automated scoring procedures for an alternative item type: figural response. Figural response items call for the completion or modification of figural material, including illustrations, diagrams, and graphs. Nineteen science items were written in cooperation with the National Assessment of Educational Progress and printed with a special ink, invisible to scanning equipment. The items were answered with pencils; response sheets were then scanned and the resulting data were processed by computer-based scoring algorithms. Implications of this technology for the future of large-scale testing are discussed.