The danger that teachers will be judged deficient just because their students have a track record of “slow” is real. But there is an answer.

Instead of judging a teacher, or a school, by the results alone, judge them by the difference between the progress the same students made last year, and this year’s progress. Or judge them by some model that takes into account everything that’s known about each of their students (do they already know English, what sort of home life do they have, any standardized test results, parental occupation, just every last thing that’s correlated with student progress) and spits out an expected amount of progress. We know a lot about modeling and validation; this can be done. (It can even be done without breaching student confidentiality—-no one needs to know how much Johnny might have been expected to learn—-just, averaged over the class as a whole, how did Mr. Rogers’ charges do, compared to what typically happens given his class’ incoming stats?)

Now, look at how the teacher does compared to what might reasonably have been expected. Teachers who, by that measure, outperform most of their peers, are probably good teachers even if their students don’t match Silver Springs, Maryland results.

Either approach gets around the problem that the best way to look like a good teacher is to get the good students.