Thursday, September 04, 2008

Merit pay for teachers

I was listening to a discussion on the radio yesterday about the pros and cons of merit pay for teachers. I think that much of the conversation missed the point, which is precisely the sort of thing that motivates me to post here.

The usual debate about merit pay seems to typically include the following arguments.

In favor of merit pay:
  • Not all teachers are the same; we should be paying teachers more if they are more effective, and paying less (and removing from the system) teachers who are not effective.
  • We want to attract good teachers by showing that they can advance and earn more money by being more effective.
  • We want to provide incentives for teachers to go the extra mile with their students by having rewards for it. Absent this, the motivated teacher and the teacher who does the bare minimum get paid the same, which is a strong dis-incentive to put in the extra effort.
...and against:
  • There is no strictly objective/unbiased way to measure teacher effectiveness.
  • Using test scores (or improvement in test scores) has a variety of negative unintended consequences, including teacher cheating, teaching to the test (and all of its related issues, which I won't go into here), and attracting teachers to the best students at the expense of the needy students (if overall test scores are the metric) or to the needy students at the expense of the best students (if test score improvement is the metric).
  • Classes have wide variances in student achievement, readiness to learn, socio-economic status, behavioral issues, learning disabilities, etc.
  • Merit pay hurts morale and teamwork among teachers because they are aware of differences in pay and because they are competing for a larger share of a fixed budget in what is essentially a zero-sum game (i.e., one teacher's merit pay increase decreases the remaining pool for other teachers, thus it comes out of those other teachers' pockets.)
  • No principal can spend enough time in each classroom to truly see how well the teachers are actually doing.
  • Merit pay "just doesn't work" for teaching: teaching is somehow different from other professions. (On caller on the radio show compared them to physicians and made the point that we don't have merit pay for physicians for similar reasons.)
I have been thinking about this a lot over the past few days, and I simply cannot persuade myself that the "against" arguments - though more numerous - outweigh the "for" arguments. In particular, while I think the "against" arguments raise good and important issues - not to be brushed aside - I think none are insurmountable, nor are they sufficient to dissuade me that merit pay is a good idea.

The biggest objection boils down to the lack of strictly objective/unbiased mechanisms for evaluating teacher performance without having the unintended consequences (mentioned above) of things like test scores. This objection is completely accurate, but I think it misses the point. It is letting the perfect be the enemy of the good. There isn't a workplace in the world that evaluates employee performance, where employees are professionals who must exercise professional judgment (i.e., not simply assembling widgets on an assembly line) that is strictly objective, and yet almost every such professional environment evaluates employee performance successfully, if imperfectly. If you go into any such environment - including schools - and ask employees to identify the stars and the dead weight, and you will inevitably find extreme consistency in the answers you get. This is one reason that many workplaces employ 360-degree feedback as part of the evaluation process. If the employees themselves can do this with such a high degree of consistency, then clearly evaluation is possible, even if it is imperfect.

In fact, I believe that an imperfect, subjective evaluation is superior to a strict "objective" evaluation because with no flexibility, no room for subjectivity, there is also no room to recognize outstanding performance or innovation that is outside of the bounds of what the "objective" metric measures. Stated another way: principals are professionals. We as parents make a big deal about good principals vs. bad principals, we are clearly evaluating them, and we are expecting them to make a difference in their schools and to use their judgment in doing so. One of the most critical aspects of a principal's professional judgment is to decide how the teachers in their school are doing. So we are explicitly paying a principal to evaluate teachers - why would we want to then make that evaluation meaningless by refusing to let that evaluation feed back into financial rewards?

I should also note that evaluations - including peer evaluations like this - are used quite successfully in many private sector workplaces without negatively affecting teamwork or morale, so I see no reason to suspect that it would suffer that fate in schools.

The argument against the use of test scores also misses the point for similar reasons. If classroom test scores (or improvement in scores) were the only metrics, then indeed it would be gamed by teachers and schools would suffer the unintended consequences described above. But this is where evaluation requires less mechanical (yikes - this means potentially imperfect and, gasp, subjective!) evaluation metrics. Goals should be set against situations. The goals for a teacher in a gifted students class absolutely should be different than the goals for a teacher in an inner-city classroom of troubled students who are reading 3 grades below grade level. It is crazy to suggest that simple test scores work as a metric for both environments, but they can and should be part of the picture. The former will - among other goals - be looking for a modest improvement in already good test scores, while the latter is doing great if they can improve scores to something closer to expectations.

And let me quickly dispense with the notion that merit pay "just doesn't work for teachers". First of all, this is an assertion without evidence to suggest why teachers are different, and why that unidentified difference is incompatible with pay-for-performance. And the analogy with professionals such as physicians fails because they do have merit pay: the good physicians get lots of referrals (or positions at hospitals etc.), and their business grows. The bad physicians get lawsuits.

Now just because I believe that merit pay makes sense does not mean that any merit pay plan makes sense. I believe that there are 2 ingredients to successful merit pay:
  • The objection above that principals do not spend enough time in classrooms is a true danger: to successfully have merit pay, a school must ensure that there is enough time for principals (and other teachers, parents, and even students) to observe and evaluate (and give feedback on) teacher performance. And it is critical to calibrate goals for each teacher based on that teacher's situation. We don't want to penalize teachers for taking harder challenges, nor reward teachers who take on "easy" classes. Test scores are certainly a piece of this, but must not be the only factor in evaluations.
  • Money must actually be available for superior teachers. My personal opinion on this is that it is a mistake to modify salary based on performance, primarily due to the long-term impact that a "good" evaluation year or a "bad" evaluation year can have on a teacher's subsequent pay. Rather, there should be an annual bonus pool, and evaluations should lead to bonus drawn from that pool. Each year's performance would determine that year's bonus; at the start of the next year, all teachers would be back at the same starting point, with an equal chance at that year's bonus. But whether done by salary or by bonus, school districts cannot implement merit pay without actually funding it.
I believe that if these two issues are successfully addressed, then teachers have nothing to fear and even much to gain from merit pay.

No comments: