Evaluation of the E‐rater® Scoring Engine for the Toefl® Independent and Integrated Prompts

Scoring models for the e‐rater ® system were built and evaluated for the TOEFL ® exam's independent and integrated writing prompts. Prompt‐specific and generic scoring models were built, and evaluation statistics, such as weighted kappas, Pearson correlations, standardized differences in mean scores, and correlations with external measures, were examined to evaluate the e‐rater model performance against human scores. Performance was also evaluated across different demographic subgroups. Additional analyses were performed to establish appropriate agreement thresholds between human and e‐rater scores for unusual essays and the impact of using e‐rater on operational scores. Generic e‐rater scoring models were recommended for operational use for both independent and integrated writing tasks. The two automated scoring models were recommended for operational use to produce contributory scores within a discrepancy threshold of 1.5 and 1.0 with a human score for independent and integrated prompts respectively.