||In this research, I examined change in interest over the span of a semester-long online computer science course. Students enrolled in two semester-long online computer science courses were asked to provide demographic information about themselves, then complete a variety of survey measures as they worked through the online class materials. Traditional self-report assessments of interest and perceived competence were collected after each of the four class units had ended (i.e., â€œquarterly self-reportsâ€), and affective, in-the-moment assessments of interest and confusion were collected both at random intervals while students were logged into the class webpage (i.e., â€œexperience-sampled questionnairesâ€) and after students had engaged with optional class examples and exercises (i.e., â€œevent-sampled questionnairesâ€). In addition, click-level indicators of behavioral engagement were collected from the class server, indicating the extent to which students were engaging with optional (i.e., nongraded) class materials (e.g., class exercises). Utilizing a dynamical systems analysis approach, results suggested that interest, perceived competence, and confusion functioned as attractive (i.e., homeostatic) set points, in that they were fairly stable and resistant to change over time. In addition, there was evidence from the event-sampled data that higher levels of confusion at one time point were associated with increases in interest at a later time point. Thus, confusion and interest were positively coupled (i.e., increases in confusion were uniquely associated with increases in interest). However, the associations between interest, competence, and confusion were moderated by participant sex, the presence of a utility value intervention, and the difficulty of the class unit with which the student was engaging. For example, men exhibited greater stabilities of interest over time (as assessed via event-sampled questionnaires), and women exhibited greater stabilities of confusion over time (as assessed via experience-sampled questionnaires). In addition, there was evidence that an intervention aimed at enhancing studentsâ€™ perceptions of utility value increased the degree of coupling between interest and confusion for female participants (i.e., for these participants, confusion was positively associated with changes in interest over time, and interest was negatively associated with changes in confusion over time). These results imply that men and women experience STEM courses differently.