Nevertheless, we recognise that our measure for student learning outcomes is not without limitation. Specifically, the total mark achieved by students in the semester exam may be influenced by a number of individual, time-specific, contextual factors, and thus may represent a relatively noisy measure of learning outcomes (see for example, Hartnett,Ro¨mcke, & Yap, 2004). We address this, in part, by conducting various sensitivity tests to validate our findings on the association between student usage of MarlinaLS and learning outcomes. These tests are presented and described in Section 5 of this paper.