Assessing the effectiveness of a course design in higher education is an almost unfeasible task. The practical inability to conduct randomised controlled trials in a natural setting limits teachers who want to interpret their design choices to non causal evaluation. With at least a moderate amount of common sense these evaluations could be useful in gathering insight into what works and what does not work in education. In our blended learning course we wanted to assess the role of formative assessment while also taking lecture attendance into account. There are certainly many confounds for which we cannot control. We found no effect of lecture attendance but formative assessments did predict a substantial amount of course outcome. This was probably due to not including the formative performance in the final grading and our emphasis in the course that to err is to be student. The validity of the formative assessment paves the way for diagnostic use and remedial teaching.