selection bias? Is KA randomly selecting these schools for their trials? Or is it a convenience sample? In that case the schools that elect to work with KA are not likely to be typical American schools.
They are still a subset of American schools.. if KA wins an A/B test wiher both A and B are at those schools, it is still a win. But the control had better be the same school!
Here you'll find all you need: http://www.khanacademy.org/#khan-academy-related-talks-and-i...