Stability of learner-defined engagement criteria

Investigate whether prospective elicitation of learners’ engagement definitions and the stability of those definitions across changing tasks or goals can support personalized engagement prediction.

Background

The paper explores whether learners’ own definitions of engagement—coded as cognitive, behavioral, and emotional criteria—can improve the calibration of sensor-derived engagement predictions. Although definition–score interactions improved performance in the exploratory analysis, the definitions were elicited after the viewing task, and the study did not assess whether such criteria remain stable across tasks or changing learner goals. The authors therefore leave prospective elicitation and definition stability unresolved.