Provisional thoughts on methodology:
The students have nine pieces of assessed writing over eight weeks. These range from 1 hour timed writings to a three thousand word essay. Teachers are paired. I've already secured the co-operation one pair, I'll get another pair at the staff meeting tomorrow. For the first (timed) essay, teachers have been asked to select their class's top, middle and bottom essays to pass over to their co-teacher for double marking. They are the ones I'll select for data, (two pairs, three per teacher, so that's 12 examples, times nine over the whole course, that's 108 examples of writing and marking. [Too many?]). And I'll stick with those twelve students throughout, whether or not they lose their high/middle/low status. That should give the data continuity. And the marking of four teachers will also guard against anomalies.
Four teachers and twelve learners will also be good numbers for other data: interviews and questionnaires - I need to seek consent of all concerned at this early stage.