A group of the nation's top education economists (Eric Hanushek, John Kain, Daniel O'Brien and Steven Rivkin) are out with an inordinately complex but highly (statistically, even!) significant study that yields sweeping findings on a whole range of teacher quality issues. The study is none too short on insight related to better structuring merit pay (most school districts are getting it all wrong); the wisdom of matching teacher race with student race (seems to be a good idea); the teacher attributes that correlate with teacher effectiveness (practically none); the damage to educational progress inflicted by first-year teachers (more significant than anything else within a school's control); and, running smack up against previous studies, a finding that inner-city school districts are not bleeding their most talented teachers, but are instead losing teachers who are about as effective as the ones who stay.
States and districts considering merit pay should take notice that districts basing teacher bonuses on their schools' overall performance growth (and not the performance of individual teachers) end up overlooking- and therefore not rewarding- the true sources of student gains: the individual teachers within any given school who really pack a punch. In other words, the greatest variations in teacher performance are found among teachers within the same school building, not between schools district-wide.
In fact, school-wide bonus plans have all sorts of fairness and accuracy problems. Small schools end up looking like they are making more progress than they do, meaning a teacher is more statistically likely to get a bonus at a small school regardless of true performance. And, because a good plan should be based on multi-year performance (no matter whether it is targeted at schools or individual teachers), the performance of current teachers gets confused with that of their predecessors, making it more likely that the wrong schools are getting compensated and deflating teachers' incentive to improve.
Actually getting through this study is hard work, and not for the faint of heart. However, some of the report's details could (in our opinion) lend support to somewhat different conclusions- a good reason to dive in and read the whole thing for yourself.