Developing a Stealth Assessment System Using a Continuous Conjunctive Model

Abstract

Integrating learning analytics in digital game-based learning has gained popularity in recent decades. The interactive nature of educational games creates an ideal environment for learning analytics data collection. However, past research has limited success in producing accessible and effective assessments using game learning analytics. In this study, a mathematics educational game called The Nomads was designed and developed to train learners’ adaptive expertise in rational number arithmetic. Players’ game log data were captured and fitted to a cognitive diagnostic model (CDM) — CCM (continuous conjunctive model). CCM lends itself well to the complex and dynamic nature of game learning analytics. Unlike traditional CDMs, CCM generates parameters at an attribute level and offers more parsimonious diagnoses using continuous variables. The findings suggest that learners’ attribute mastery improved during the gameplay and that learners benefit from using the scaffolds for three of the attributes instructed by the game. This study presents the application of a powerful new tool for game learning analytics. Future studies can benefit from more generalized analytics models and more specified learning attributes and game tasks.