Modeling the effects of mastery measurements in a digital formative assessment system with a Bayesian network

Abstract

Mastery measurements are short, criterion-referenced formative tests for assessing student performance in hierarchically-structured learning domains. Although many digital learning materials have built-in mastery measurements, few systems have been subject to empirical research. This study investigated the effects of mastery measurements on feedback usage, learning activities and learning progression within a Moodle course on grammar and spelling. Moodle log-files and scores from 407 students in 18 classrooms (Grades 6–9) were analyzed on the level of test-retest-sequences. A Bayesian network approach was used to model dependencies within the sequences and characteristics of the students and context. Data analyses revealed a pattern of feedback and learning activities that increased the probability of learning progress within a test-retest-sequence.