Eye movements during code debugging:expertise differences and didactic gaze

A recent development in programming education is the popularity of tutorial videos that show a screencast of a (expert) teacher who explains how to perform a programming task. However, these videos are often visually cluttered and following the expert’s reasoning is presumably difficult for novice learners. One approach to foster attention guidance of novices in such videos is Eye Movement Modeling Examples (EMMEs: Van Gog, Jarodzka, Scheiter, Gerjets, & Paas, 2009). EMMEs superimpose visualizations of a didactically behaving person’s eye movements onto instructional videos. However, little is known about the characteristics of experts’ naturally occurring gaze patterns in comparison to their didactic gaze patterns while creating EMMEs. To investigate these gaze characteristics, a group of experts (N = 20) first debugged short computer code snippets while their eye movements were recorded. Subsequently, the experts were instructed to explain their solution didactically in an EMME video. A group of novices (N = 18) debugged the codes without the subsequent instruction to behave didactically. Our first research question is how experts’ natural gaze patterns differ from those of novices while debugging. Based on previous expertise research we investigate differences in gaze patterns between experts and novices. Our second research question is how experts’ gaze patterns change when being instructed to behave didactically. This explorative investigation provides insights into the characteristics of didactic gaze. Later, this knowledge could be used to create more effective EMMEs in programming education