Measuring Differences among Non-Randomized Groups: An Epidemiological Model for Identifying Successful School Programs