High school teachers--they apparently flip every other year.
They were damned good teachers.
There was a hitch, though. When evaluating the lot of them, it was a a lot easier to high-rank those dealing with advanced kids and not on-level kids or even doing remediation.
Those with nifty tech also had an easier time. They could pivot from running the lesson to kicking the center of attention to the students, or sideline the students productively while dealing with some sub-group that needed attention. Those with paper-and-pencil classrooms couldn't just pull up alternatives or walk to another part of the white wall (instead of white board) or assign control over classroom participation to a lead student in a 1-to-1 classroom with a couple of taps on an interactive white board while bringing up stuff to address some other issue for a subgroup of students.
And, more important than tech, those not in Podunkville tended to have a lot more community and professional engagement than those in a school with 500 kids over 4 grades 60 miles from the nearest fire hydrant. But we didn't see a way of addressing this issue, because you can't judge what you can't see.
Those all aligned. If you're in a school of 3000 it's more likely advanced classes will "make", there'll be funding for more classroom tech, and you're likely to be in a population center where there are lots of possibilities for engaging in the wider community (whether it be local or educational).
Note that the PAEMST folk are selected by a group of teachers in each state. The applications are reviewed by teachers, the reviewer scores forwarded to teachers, and the teachers provide feedback to the PAEMST nominees for how to improve their application prior to being forwarded to DC.