Why Doesn’t Merit Pay Work?

Dick Startz this morning discusses one of many experiments indicating that performance pay for teachers not only fails to improve student outcomes, but also doesn’t do much to change teacher behavior. I don’t think people are sufficiently puzzled about why that is, so I’m glad Startz offers some possible explanations:

  • A $3,000 bonus is nice, but it’s something under five percent of a NYC teacher’s salary. Maybe that’s not enough to make a big difference.
  • The average NYC school has 60 teachers. Perhaps knowing that the fruits of your extra labor are going to be split 60 ways means your incentive is very small.
  • The program gave one year bonuses, not permanent raises. Could be that also meant teachers felt it wasn’t worth making big changes.
  • The program was not accompanied by new tools to help teachers do better.
  • Nothing in the program changed who taught in the school. It was focused only on changing behavior of the existing teachers.

Those are all certainly possible factors, but they largely sidestep the central conundrum. After all, even a bonus that is small on average should – under simple models of human motivation – change behavior at the margin. Yes, for many teachers the incentive may be too small to justify changing their behavior, but there should still be a sizable number of teachers for whom $3,000 is motivating.

Of course, human motivation is somewhat messy and irrational. A year is a long time to delay an incentive, for example, and people will (irrationally) discount its value accordingly. Still, even an incentive that is inadequate on average should motivate some teachers to take at least the relatively easy steps available to them to earn it. Something else, I think, must be discounting the value of the incentive even further.

So it seems to me that part of the answer has to be that teachers are for the most part not very certain about how to improve student outcomes. The expected value of a reward will depend on the probability that attempting to earn it will be successful. In teaching, that sort of probability can be very difficult to judge.

As a teacher I often have a pretty good sense for the extent to which my students “get” the content we’re working on. Enough of my formal and informal assessments remain the same from year to year that I am also usually fairly confident about how each cohort stacks up against previous years’ students.

What’s much harder is explaining that student success (or lack thereof). A lot of factors determine how well my students do on, say, a test, so it’s hard for me to know with any certainty whether they’re doing better (or worse) because of any particular activity or classroom practice. And to the extent that I am certain about any particular teaching practice, I’m already trying to employ those that I’m most sure of. Expanding my efforts means, almost by definition, that I’m going to start trying things I’m less confident about.

No matter what your policy preferences are, this should concern you. If you tend to favor merit pay schemes, it should worry you that so many programs spend significant amounts of money but produce no results. At the same time, even if you find the idea of merit pay repellent it should be troubling that teachers are apparently either unwilling or unable to move the dial further on student achievement even when offered a reward.

I don’t think that’s a big knock on teachers, but it probably is a cause for educational pessimism.

This entry was posted in Teacher Compensation and tagged , . Bookmark the permalink. Post a comment or leave a trackback: Trackback URL.

One Trackback

Leave a Reply