BackHome
Refresh: OffView Personal Schedule
Individual Presentation Summary
Share...

Direct link:

Fair or Flawed? Analyzing Biases in Automatic Text-Response Scoring

Sun, April 14, 7:45 to 9:15am, Convention Center, Floor: First, 121C

Abstract

To measure the fairness of automatic scoring, models were trained on text-responses from the PISA 2015 reading comprehension tasks and results were compared between students of different genders and students who speak a different language at home. Scoring models systematically overestimate the performance of non-native speakers, giving them an advantage.

Authors