A human-centered AI-augmented approach to contextualizing test performance on large-scale assessments

• In Event: Innovation Demonstration I

Abstract

Process data, collected from large-scale assessments, offer opportunities to explore students' test-taking processes, but manually labeling them is hard to scale. We proposed a human-centered AI approach to make sense of process data to contextualize performance, and to help identify sources of low performance and generate rich feedback.

Authors

• Hongwen Guo, Hongwen Guo/ETS

• Matthew S Johnson, ETS

• Luis Saldivia, ETS

• Michelle Worthington, ETS

• Kadriye Ercikan, EDUCATIONAL TESTING SERVICE