The Math of Merit Pay

Bured in yesterday's Ed Week write-up ($) of McCain's speech was a quote that nicely sums up the common concerns over merit pay:

Bettye Oldham, a retired teacher from Cincinnati, said she had mixed feelings about the merit-pay proposal.

“I think it’s a good idea if it’s run correctly,” she said. “But you can’t judge teachers the way you can judge manufacturing companies.” Some students come to school with “issues that prevent them from learning,” she said, which can make it harder for teachers to reach them.

I picture what Ms. Oldham might look like, and I can see her mulling the pros and cons of merit pay. She's probably worried that it would reduce teachers to nothing more than cogs on an assembly line, but also understands the implicit logic that good doctors and lawyers make more money than bad ones. Teachers, on the other hand, are paid only by credentials--all teachers in the same district with a bachelor's degree and 12 years experience, for instance, are paid the exact same amount no matter their talent or specialty area.

Merit pay seeks to measure some of the work teachers do. While the precise math is a little too complicated for this blog, in essence it looks like a fairly straightforward algebra problem:

Student achievement = age + prior achievement + student demographic information + teacher quality + measurement error

A common argument against merit pay, and the one repeated by Ms. Oldham, is that there are too many human elements for the measurement of teaching. But that argument overlooks even the simple version of the merit pay equation. Race, income and prior achievement are all factored in, and the error term captures all those things that are not predictable--a student's situation at home, whether they had a good breakfast, whether it's too hot or cold, etc.--all things that surely affect the student's score but are not measurable.

It might seem a little too whiz-bang to trust the capabilities of modern statistical computer programs, but mathematicians far greater than I are working to refine the exact statistical equations. We've made giant strides in our ability to compute the value an individual teacher adds to student learning. Let's not let fear of complex, modern math get in the way of promising reform.

Update: Fred Klonsky responds with the "human element" argument. He's making my point for me though, because it really goes back to a fear of assessment of teachers' work. We are comfortable measuring a baseball player's batting average, even if they had a bad breakfast, a rough flight the night before, or an argument with their spouse. In fact, their results are published the next morning in your daily paper. I can't imagine A-Rod explaining poor results on the field with his off-field Madonnna shenanigans. There are "human elements" to everything in life.