Reply #97: I'd have to see the model to answer specifically [View All]

[Recursion]

The weights of the classroom, school, and district performance are the I coefficients; they are found by some more or less (probably less) meaningful heuristic regression over past data (this is difficult to do well for widgets, let alone students).

I'm not meaning to come across as saying this formula is a good idea; my point is just that it's well known from systems theory and has had useful applications before, even in "soft" situations (the wargames that convinced Shinseki we needed 400k troops for Iraq, for instance). In general this works better on large scales than small ones.