At a recent meeting in New York City, Secretary of Education Arne Duncan said that “we as a country don’t know” how much student test scores should count as part of teachers’ evaluation. He said it shouldn’t be zero, and it shouldn’t be 100%. But it should be somewhere in between. As to what the number should be, the secretary said, “we don’t know.”

Here’s a thought: What if the current methods of calculating value-added are inaccurate? What if they are fundamentally flawed? What if they say nothing about teacher quality? What if they reflect who is in the class rather than teacher quality?

What if, say, a few years from now, we look back and realize they are junk science?

How much should they count then?

And if we don’t know whether they are accurate, and we don’t know if they are a reasonable measure of teacher quality, and if we have no evidence that their use in evaluation helps teachers improve or students achieve, why are we counting them at all? Shouldn’t we wait until we have clear evidence that the methods we use to evaluate teachers and principals are accurate, fair, reliable and valid, before putting them into practice?

I know that “we can’t wait,” but shouldn’t we wait long enough to know that what we are doing will help and not harm?

Or, are we still building a plane in mid-air?