[MrMon]

This is why as I wrote my idea, I started narrowing it to eliminate the problem of different student socioeconomic backgrounds. If you could take just urban districts that tend to be pretty uniform and compare pay to student success, something might turn up. For instance, St. Louis, Detroit, Cleveland, Washington, and Philadelphia would all be pretty similar, mostly black, no real affluent neighborhoods attending, or too few to make a difference. You really can't include NYC, too many other kids in the district. LA and most of the west is too Hispanic. Find as many similar districts as you can, then compare them on teacher pay. Do something similar with sururban or rural districts as well, try to find clone districts, then compare on teacher pay. With the thousands of districts out there, surely it should be possible to find enough that are similar to come up with something that is statisically signifigant.