There is an independent experiment by Monterrey Tech from [c. 1992]. It predates C++ and TeachScheme!. They compared vast numbers of students with the AP exam at the time, with either one year of Pascal or one semester of Scheme followed by one year of Pascal. The Scheme students performed vastly better. Unfortunately they couldn't repeat this particular result with different instructors, and so the university concluded that it was just one teacher's enthusiasm.