Automated Scoring of Scientific Practices Using the Next Generation Science Standards Learning Assessment System

• In Event: Poster Session 1
  In Poster Session: Issues in Assessment Development and Practice in Public School Classrooms

Abstract

Assessment within the American science classroom is changing. Due to implementation of the Next Generation Science Standards across the US, scientific practices are given equal priority to scientific content. However there are few assessments that explicitly measure a scientific practice and even fewer assess multiple practices simultaneously. The Next Generation Science Learning (NGSL) assessment system uses open-ended prompts to assess student’s competencies of data analysis, explanation construction and argumentation. The use of open-ended assessments is labor intensive. This study explores the potential of using natural language processing and machine learning R packages to generate automated scoring models. The automated scoring of open-ended responses greatly increases the usability assessment systems like the NGSL for both teachers and researchers.

Authors

• A.J. Womack, University of Missouri

• Eric Wulff, University of Missouri - Columbia

• Troy D. Sadler, University of Missouri - Columbia