Upsetting the Apple CART: Classification Trees for Process Data from Performance-Based Assessments

• In Event: Process Data and Data Quality

Abstract

This research analyzes process data from a virtual performance-based assessment (VPBA), where examinees manage a simulated patient scenario. Through the application of classification trees to sequenced actions, this study illuminates an interpretable, data-driven methodology for highlighting salient patterns in VPBAs, enhancing understanding of examinee actions within complex action spaces.

Authors

• Nate Smith

• Rich Feinberg, National Board of Medical Examiners

• Chunyan Liu, NATIONAL BOARD OF MEDICAL EXAMINERS

• Lisa Keller, UNIVERSITY OF MASSACHUSETTS