The Case of Self-transitions in Affective Dynamics

摘要: Affect dynamics, the study of how affect develops and manifests over course learning, has become a popular area research in learning analytics. Despite some shared metrics questions, researchers this have differences they pre-process data for analysis [17]. Specifically, differ treat cases where student remains same affective state two successive observations, referred to as self-transitions. While most include these their data, D’Mello others argued last few years that should be removed prior analysis. choice reflects intended focus paradigm on transitions out an state, difference preprocessing changes meaning metric used. For around decade, community used L evaluate probability affect. is largely believed value 0 when transition at chance, true original use metric. However, paper provides mathematical evidence does not chance if self-transitions are removed. This shift problematic because previously published statistical analyses comparing values wrong value, incorrectly producing lowered p many reporting significantly more likely than actually less frequent.