In this paper, we argue that, where we measure student attendance, this creates an extrinsic motivator in the form of a reward for (apparent) engagement and can thus lead to undesirable behaviour and outcomes. We go on to consider a number of other mechanisms to assess or encourage student engagement -- such as interactions with a learning environment -- and whether these are more benign in their impact on student behaviour i.e. they encourage the desired impact as they are not considered threatening, unlike the penalties associated with non-attendance. We consider a case study in Computer Science to investigate student behaviour, assessing different metrics for student engagement, such as the use of source control commits and how this measure of engagement differs from attendance.