Landing : Athabascau University

StudentLife: Assessing Mental Health, Academic Performance and Behavioral Trends of College Students using Smartphones

http://studentlife.cs.dartmouth.edu/studentlife.pdf

A totally fascinating study of students conducted using a massive amount of automatically collected data from smartphones along with other data collected from other systems and via surveys to come up with a large set of correlations relating to everything from mood to GPA. This would win a top paper award in any conference I can think of.

Too much to summarize here, and many more questions emerging from it than it answers, but this should keep a load of researchers busy for years to come. I'm certainly going to be picking this over carefully now that I've read it through once. I highly recommend that anyone involved in education (staff or students) should read this! But it should be read with great care and with all critical faculties on full alert. This was a very specific group of students in a very specific context and it would be highly dangerous and irresponsible to extrapolate any generalizations at all from any of this, though I bet some people will. There are lots of things that warrant further investigation - active students were happier and did better but lack of activity, especially at night, seems correlated with higher GPAs, for example, and there are some big fuzzy areas in the sampling that involved a lot of interpretation that was unlikely to be particularly accurate much of the time. The finding that I find particularly appealing is the discovery that classroom attendance had no correlation with academic performance at all: I almost laughed out loud at this one. As always, however it's not what but how that matters. This suggests to me that someone really needs to work on their classroom activities rather than that classroom teaching does no good, and I would really like to know a lot more about the students who skipped classes before even drawing conclusions from this small dataset let alone more broadly. The other big issues here surround the need for careful interpretation and more qualitative data to explore causes: all this shows are correlations, some of which seem to imply obvious things (e.g. students that study rather than party tend to get better grades but they tend to be lonelier) but many of which are more complex and should be considered in context and at a whole systems level.

The anonymized dataset is available for downloading.