The Quantified Student: An App That Predicts GPA

It sees you when you're sleeping ... it knows when you're awake ... it knows if you've been hitting the books, so be good for goodness' sake!

No, it's not Santa Claus. It's the digital Jiminy Cricket each of us carries in our pocket, otherwise known as a smartphone.

In a small experiment, researchers at Dartmouth College have shown that data automatically collected by an Android app can guess how students are spending their time — predicting their end-of-term grades with scary accuracy.

How did they do this? And what does it mean?

The how is simple, although even the lead investigator, Andrew Campbell, admits it sounds a bit "creepy." A smartphone generally has Wi-fi, GPS to detect location, an accelerometer which detects motion and a microphone, which can pick up nearby sound. The phone also senses whether or not it's being charged or being used.

Using this information and a map of the campus, the researchers designed an app that modeled several different behavior scenarios:

1) Sleeping: It's nighttime, your phone is charging in your dorm room, and you are not interacting with it.

2) Physical activity: You're walking or running.

3) Studying: You're in the library, a computer lab or a coffee shop that is a popular study location. The noise level is relatively low. You're there at least 20 minutes. The app infers your level of focus by seeing how often you check the phone during the study period, and whether you're staying in one spot or roaming around.

4) Partying: You're at a fraternity or sorority house. You don't live there. It's loud. It's probably a Wednesday, Friday or Saturday night, the biggest party nights for Dartmouth students.

And the results? You may be flabbergasted to learn that students who spend more time studying get better grades.

Slightly more surprising, students tended to perform better when they buckled down towards the end of the semester. After the midterm, "A" students partied less, stayed at home more, and spent less time in conversation. But what's interesting is that this relationship held true whether they started out as relative extroverts or introverts. It wasn't the absolute time spent partying, in other words, it was the ability to prioritize that really counted.

Also of note: students with better grades studied in louder locations. Were they benefiting from study groups? Maybe.

This experiment included only 30 students, and being Dartmouth undergrads, their grades were all extremely high, ranging from a B to an A average. The next step, says Campbell, is to test it on a larger, more diverse set of subjects, this time at the University of Texas at Austin.

If the results hold, a natural next question is, Who might find these patterns useful, and for what?

Campbell, a computer scientist, has a longstanding interest in what he calls "persuasive technology." He dreams of making this app available in the app store to help students improve their behaviors and in turn, classroom performance — a Fitbit for your brain.

"Can we build an instrument that is reliable? And if we can, how do we provide that feedback to a student and do a classic intervention?"

But what about privacy? "It's the Achilles heel of all these applications," Campbell admits. Assuming this technology takes hold, should professors, deans, advisers get access to student tracking data? What about parents? For what purposes and under what circumstances, and who decides? "It's sensitive data," says Campbell. But in the right hands, he argues, it could be empowering.