To transform education, Donorschoose hires a data scientist
FORTUNE — What kind of tech do teachers really need? Depends on whom you ask.
There are plenty of companies and causes fighting for more laptops and tablets and smartboards in public school classrooms. But on Donorschoose.org, where a quarter of a million teachers hailing from roughly half of the nation’s public schools have logged requests for their classrooms, the most frequently requested item is an inexpensive next generation overhead projector called a document camera.
These types of findings have been bubbling up since April of 2000 when Charles Best, a former Bronx history teacher, started Donorschoose to help donors and teachers connect directly online. Now Best plans to use a dozen years’ worth of data to advocate for those educators. After all, teachers often turn to Donorschoose to request help getting the tools and supplies they need most. In the grand — and often political — struggle to identify what schools need, Best believes Donorschoose can help policy makers who control government spending listen to teachers. Says Best, “Laptops are important, but before you spend a million dollars per school providing one laptop per child…won’t you please spend $5,000 per school equipping every classroom with a document camera?”
His goal is nothing short of audacious: Best aims to influence $1 billion in education spending over the next five years. “It’s not meant to be partisan,” says Best. “Our only political stance is this: listen to these teachers.”
The idea for harnessing this data began when Donorschoose held a hacking contest in 2011. Among the most intriguing explorations of the data was a project done by Lisa Zhang, a Canadian undergraduate who used Donorschoose data to look at, among other things, the influence of a teacher’s gender on the types of projects that were submitted and got funded. Some stereotypes were underlined: Male teachers were more likely to express anger in their project requests, for example. They were also slightly more likely to be funded; the researcher attributed this to the anger, which was often righteous anger against injustice. There were also humorous findings: Married female teachers were most likely to use vaguely sexual terms in their requests. (“We’re going to penetrate the subject material.”) Best and the Donorschoose team began to wonder what else they could discover by applying rigorous research to the data they’d collected. Like many of their peers in big business, they decided to hire a data scientist.
With donors and board members that include LinkedIn LNKD CEO Jeff Weiner and Union Square Ventures managing partner Fred Wilson, Donorschoose has always been run more like a Silicon Alley startup than a traditional nonprofit. “We’re very data driven, but the type of data that drives our day-to-day is our web analytics dashboard,” says Chief Technology Officer Oliver Hurst-Hiller, who himself arrived at the company from Microsoft MSFT where he’d managed product engineering products for Bing. “A data scientist uses advanced analysis to find something you didn’t even know you were looking for.”
Best and his team spent nearly a year searching for the perfect candidate. Data scientists are tough to land — both because they are so highly in demand and because the set of analytical skills paired with intuition and the ability to digitally spelunk is uncommon. They got help with the hiring process from OkCupid data scientist Max Shron, who often held up as a nerdtastic rock star among his peers for his work on the OkTrends Blog. Ultimately Shron encouraged them to bypass two more credentialed candidates in favor of Jay Garlapati, a 22 year-old University of Chicago mathematics major who had studied evolutionary game theory. The catch? They had to wait until he graduated to hire him.
Garlapati officially started at DonorsChoose on July 1. When he applied, he hadn’t realized the outfit was a nonprofit. “I was looking for an organization that wasn’t just messing with data, but really asking meaningful questions and trying to make a difference,” says Garlapati. And he adds, bashfully, “I also hadn’t really had a job.” Because Garlapati’s work holds the potential of expanding Best’s vision for DonorsChoose, Best turned to Weiner along with Reid Hoffman and investor Anthony Meyer to underwrite the position. That way, the organization can ask Garlapati to spend half his time “figuring out how to change the world,” as Best says, without spending donor dollars.
First up, Garlapati has spent much of the last few months studying value-added scores, a score used to assess teacher efficacy by tracking students’ performance over time and measuring how much they improve under a particular teacher’s instruction. Says Best, “As a result of never being able to prove that teachers on our site tend to post better than average test scores gains, major channels of funding were shut off to us.” Best wanted analytical proof for his core belief that teachers using Donorschoose are considered above average in their practice. So far, the data has born out.