Modeling Complex Data From Simulations to Assess Middle School Students' Next Generation Science Standards–Aligned Science Learning

• In Event: Novel Developments in the Use of Response Process Data in Assessment

Abstract

Although NGSS emphasizes the integration of science practices and content knowledge, measuring how students participate in such learning is challenging. This study explores how complex data from simulations can measure how 8th-grade students engage in science practices to understand chemistry concepts. The data were fit to six Item Response Theory (IRT) and Bayesian Network (BN) models and compared in terms of fit and possible interpretations. The results suggest that the multidimensional IRT model had the best relative fit and provided estimates for different dimensions. The Dynamic BN model provided unique information about student learning based on the order of actions in the simulation. The study shows the potential of leveraging process data from simulations to assess students’ NGSS-aligned learning.

Authors

• Emily Toutkoushian, The American Board of Anesthesiology

• Kihyun "Kelly" Ryoo, University of North Carolina - Chapel Hill