What behaviours would we take as evidence that students are cheating? This article detects groups of students working together though the similarity of their answers and the proximity in time that these answers were submitted. It then applies analytics to their online behaviour to see if any other sorts of analytics can catch the unauthorized collaborations. They get good grades, but that doesn't set them apart. Nor do their interactions with the online course. Nor does the timing of their activities. Indeed, as the authors recognize at the end of the paper, "we have no hard proof (like video feed) that students are performing such academic dishonesty together." This should be taken, I think, as a cautionary tale. How much of our honesty-algorithms are being based on our presumptions about cheating behaviour, and how much is supported by hard evidence? And what are out algorithms learning as a result?
Today: 4 Total: 4 [Direct link] [Share]
