Merit Pay: It Won't Work

Diane Ravitch on merit pay:

What's Wrong With Merit Pay

Dear Deborah,

Over time we have developed a very solid and smart community of readers who like to argue with us and with each other. That is as it should be. And of course we need to bridge differences—or disagree—with them, too, as we do with each other.

So the subject today is merit pay. This is an important topic because it has become clear that President Obama has decided to hang his hat on this idea. It has not yet been explained just what he means by merit pay. Does he mean that teachers should be paid more for teaching in what is euphemistically called “hard-to-staff” schools? Or paid more for teaching in areas where there are shortages, like certain kinds of special education or subjects such as math and science? Or paid more for mentoring other teachers? Or paid more for teaching longer days?

I would call such compensation “performance pay,” rather than “merit pay,” because teachers are paid more for doing more.

But I have a feeling that what the Obama administration has in mind is paying teachers more based on their students’ “value-added” test scores. So if their students see increases in their scores, they will get “merit pay” to reward their supposedly superior teaching.

I believe that this is the direction the administration is heading and that it is the purpose of the millions that will be spent on data warehouses in every state. And it is why Secretary Duncan has told the governors that they will get their stimulus money only if they collect and report data to the U.S. This was an odd request because some of the data he asked for is already available, such as the gap between state and NAEP scores (previously published in Education Week, for example, and no secret).

There are several reasons why it is a bad idea to pay teachers extra for raising student test scores:

• First, it will create an incentive for teachers to teach only what is on the tests of reading and math. This will narrow the curriculum to only the subjects tested.
• Second, it will encourage not only teaching to the test, but gaming the system (by such mechanisms as excluding low-performing students) and outright cheating.
• Third, it ignores a wealth of studies that show that student test scores are subject to statistical errors, measurement errors, and random errors, and that the “noise” in these scores is multiplied when used to make high-stakes personnel decisions.
• Fourth, it ignores the fact that most teachers in a school are not eligible for “merit” bonuses, only those who teach reading and math and only those for whom scores can be obtained in a previous year.
• It ignores the fact that many factors play a role in student test scores, including student ability, student motivation, family support (or lack thereof), the weather, distractions on testing day, etc.
• It ignores the fact that tests must be given at the beginning and the end of the year, not mid-year as is now the practice in many states. Otherwise, which teacher gets "credit," and a bonus for score gains, the one who taught the student in the spring of the previous year or the one who taught her in the fall?

I believe that our readers are right when they predict that merit pay of the stupidest kind is coming. I predict that it will do nothing to improve our schools. A few weeks ago, the conservative Manhattan Institute released a study showing that merit pay had no impact on test scores in 200 schools in New York City that are trying it. In fact, scores went down in larger schools that offered bonuses. This little experiment in schoolwide bonuses is costing taxpayers $20 million a year.

Now it is possible that scores may go up in later years; this is only the first year, after all. But what is most interesting is the subdued release of this study. When the Manhattan Institute releases a study, it often holds a press conference to announce the results. This study, however, had no fanfare; its study was quietly posted on MI's Web site; no press conference, no press release. Somehow I suspect that the study would have been released with bells and whistles if the scores had flown upward.

Here is my prediction: Merit pay of the kind I have described will not make education better, even if scores go up next year or the year after. Instead, it will make education worse, not only because some of the "gains" will be based on cheating and gaming the system, but because they will be obtained by scanting attention to history, geography, civics, the arts, science, literature, foreign languages, and all the other studies that are needed to develop smarter individuals, better citizens, and people who are prepared for the knowledge-based economy of the 21st Century. Nor will it identify better teachers; instead, it will reward those who use their time for low-level test preparation.

Is it possible to have an education system that mis-educates students while raising their test scores? Yes, I think it is. We may soon prove it.

Diane