Effect of Immediate Feedback and Revision on Psychometric Properties of Open-Ended GRE Subject Test Items GRE NLP
Powers, Donald E.
- Publication Year:
- Report Number:
ETS Research Report
- Document Type:
- Page Count:
- Subject/Key Words:
Graduate Record Examination (GRE),
Constructed Response Items,
Natural Language Processing (NLP),
Automated Scoring and Natural Language Processing,
Automated Scoring of Written Content
Registered examinees for the GRE Subject Tests in Biology and Psychology participated in a Web-based experiment where they answered open-ended questions that required a short answer of 1-3 sentences. Responses were automatically scored by natural language processing methods (the c-rater scoring engine) immediately after participants submitted their responses. Based on natural language processing methods (the c-rater scoring engine), participants received immediate feedback on the correctness of their answers and an opportunity to revise their answers. A significant revision effect was found. Participants were able to correct many of their initial incorrect answers, resulting in higher revised scores. In addition, the reliability of revised scores was higher than initial scores, although the correlations of the initial and revised scores with the GRE Subject Test scores were similar.