To identify and better understand the experience of these teachers, we started by studying 90,000 teachers across four large, geographically diverse urban school districts. We also examined student academic growth data or value-added results for approximately 20,000 of those teachers. While these measures cannot provide a complete picture of a teacher’s performance or ability on their own–and shouldn’t be the only measure used in real- world teacher evaluations–they are the most practical way to identify trends in a study of this scale, and research has demonstrated that they show a relationship to other performance measures, such as classroom observations.3 We used the data to identify teachers who performed exceptionally well (by helping students make much more academic progress than expected), and to see how their experiences and opinions about their work differed from other teachers’–particularly teachers whose performance was exceptionally poor.