Creating a new generation of data to solve social problems.
When “Tara” dropped out of high school to have her baby, few of us were truly shocked. All of her teachers had seen the warning signs along the way: her troubled home environment, her slipping grades, her physical changes, and, of course, the increasing teasing she endured. We could see with 20/20 hindsight the road that led to that day.
It was so easy in retrospect to diagnose what went wrong, but surely there could have been a way to avoid this, we reasoned. While Tara was far down her path toward dropping out when she became my geometry student, the symptoms must have been there in middle school or even earlier.
As I looked back at the students we lost—to juvenile detention, pregnancy, or simple academic failure—I couldn’t help but wonder if we were missing something. There had to be some kind of tool to help figure out whether students were on track for dropping out before they were at imminent risk.
It turned out there was a model for that kind of risk prediction. But it was in the most unlikely of places.