Job evaluations are a private matter. But for teachers, evaluations have become part of an acrimonious public debate. The debate centers on student test scores, and to it I want to add a simple message. Yes, please evaluate teachers based on how much students improve at tests. But only do it if you can produce significant results that hold consistent over more than one school year. When you can't, please don't.
I'm hardly a professional statistician. But I do teach math, and I remember enough from statistics to know that the bar for proving statistical significance is pretty high. It depends on stuff like confidence intervals, but one would expect a large majority of teachers to produce improvements to student test results that, on average, are not significantly different from some teacher fished from the sea of teachers at random.
That's not to say that numbers shouldn't play a role in evaluations. But it is to say that, for most teachers, the data won't tell you anything very reliable. Certainly, the compulsion among education reformers to rank every single teacher on a "value-added" scale seems a dubious statistical enterprise at best -- and at worst a mean-spirited fantasy.
Rather than focusing on all teachers, the data debate ought to center squarely on the few teachers at tails of the famous bell curve. For the super high performers consistently out at the right-hand tail, I say pay these people more -- a lot more -- perhaps several tens of thousands of dollars more. Meanwhile, the super low performers consistently at the left hand tail ought to get out of teaching.
What about the big majority of teachers bouncing around at the center of the curve -- the group for which test score data really shouldn't be used as a basis for decision making? I'm a parent as well as a teacher, and as a parent my desires are modest. I merely want the best teacher in the world, year after year. Still, I will gratefully accept a teacher who works hard, cares about my child and is competent in their subject matter. It's unrealistic to expect more, except perhaps that my child's teacher keeps pushing to get better. We need to have an evaluation system that fosters that development process, not one centered on a numbers fantasy.