Kimberley Arnold’s article gives a very readable overview of the Signals project carried out at Purdue University to apply analytics to improve student success. Their approach is interesting in that the information yielded by their application of analytics is fed back at individual student level, and at least some of the onus on recovery and/or improvement is given to the student.
If the stats are sound and the conclusions drawn are defensible, then it seems to me that this is a pretty good way to go. The Signals project is identifying students who may be at risk, based on their personal data and study behaviours and alerting both them and their tutors/faculty support to that possibility.
Each student has a real time view of a traffic signal categorising them as on course for success (green), at moderate risk (amber) or high risk (red).
The student can access links to positive actions that can help them to get back on track and tutors/faculty support staff can opt to intervene directly.
What wasn’t completely clear to me was the extent to which tutors are responding to the signals and offering intervention support, as opposed to the students’ own response to the visual cue offered by the signal. If the former, then the Signals project reflects other ongoing approaches which offer positive support driven largely by the institution. If the latter, and the responsibility for refocusing efforts lies mainly with the student, then this perhaps alleviates some of the ethical concerns associated with a learning analytics approach. We’re simply putting the information out there and allowing students, as more informed individuals, to choose how (and whether) to respond. Much as I like the sound of this latter approach, my experiences are that many of the students who require proactive intervention do not tend to seek it themselves – for a whole range of reasons often tied up with the cause of their falling behind in the first place. The visibility of this system though brings possibilities of a more balanced view of exploiting analytics from a series of mutual perspectives, and that can only be good.