Session Summary
Share...

Direct link:

Applying Machine Learning in Next-Generation Science Assessment

Sat, April 18, 10:35am to 12:05pm, Virtual Room

Session Type: Symposium

Abstract

Developing next generation science assessments to measure students’ ability to undertake scientific practices while drawing disciplinary core ideas and connected them with crosscutting concepts is challenging. Several federal-funded projects have made efforts to meet the challenges. This session brings together four such projects that have individually and significantly advanced our understanding and practice of developing next generation science assessments. While demonstrating principles and rationales of assessment development, presenters focus on applying machine learning to automatically score constructed response items in order to make the assessments accessible to teachers for classroom uses. The studies present empirical evidence to support the promises and essential of using machine learning and point out the challenges of developing assessments that are scorable by machine learning.

Sub Unit

Chair

Papers

Discussant