Incentive works. Want people to do something better? Offer to give them something they want, like money. This simple theory has driven the American economic engine for centuries. In fact, the financial crisis was partially triggered by incentives becoming misaligned with long-term risk. They matter. One industry where incentive is clearly absent, however, is education. Recently, people have begun realizing this, and merit pay for teachers has become a heated debate.
Teachers unions are vehemently against incentive-based pay. One problem, they say, is that test scores aren't a fair indicator of how much a child has learned. There's also the worry that students will just be taught how to do well on the test, instead of receiving a broader education. Another problem is that different students have different inherent learning curves, so teachers can't be evaluated fairly against each other, since their students may have different aptitudes. These are fair criticisms.
So why not devise an evaluation methodology that answers such complaints? It might not be as impossible as it seems.
First, you need to set the baseline for each student. You could do this through IQ tests, determining reading and math comprehension levels, identifying learning disabilities like dyslexia, attention deficit disorder, etc. Such information would be on record, so these variances could be taken into consideration when a teacher's students are evaluated.
Then, rather than just standardized tests, you could have a more robust way to evaluate students. Each semester, school administrators could meet with a random sample of each teacher's students for a short time. They could ask the students questions that would qualitatively and quantitatively demonstrate the teacher's performance.
For example, on the qualitative level, students could be asked what the most memorable history lesson was from the term. Or maybe to explain five things they've learned that semester in biology. The administrator could have a fifth-grader walk them through a long-division problem step-by-step. You could even ask questions that help determine parental involvement, to help establish the student's baseline.
Then, you could have a more quantitative testing component. On some level pencil and paper are unavoidable. If students expect to succeed in life, they need to be able to perform adequately on a test -- but the exams don't have to be standardized. They can include items like a short essay, a math problem where you can see their work, etc. This pencil and paper testing portion can count for as little as 50% of the teacher's overall evaluation. And remember, a student's aptitude is already taken into account, so the test would be graded on a curve accordingly.
Obviously, such a methodology wouldn't be perfect -- but neither is any incentive scheme in any industry whatsoever. Even if you're in sales, and paid on commission, you might have better luck with certain clients by the luck of a draw. There are times when some employees appear better than they should and others don't appear as stellar as they are. But on an aggregate basis, such irregularities should even out. And incentive pay would likely be based on a sort of bell curve anyway, so the teachers who really shine and those who are really terrible should be relatively easy to determine if the evaluations come out consistently for a few years.
Of course, there is cost to consider. This would likely involve the hiring of new administrators -- but maybe not as many as you think. Let's do a little math.
Imagine a school with 2000 students with a student-teacher ratio of 20, meaning 100 teachers. If each student has seven courses, and each teacher has a planning period, that comes out to around 140 students per teacher. Take a random sample of those -- say 20 per semester (40 per year). The evaluator spends 30 minutes with each student in the qualitative section and uses another 30 minutes to grade the written test and compile the full results. That would require 2000 working hours per semester for the school's evaluation process. If you add 5 employees for this evaluation role -- that's 400 hours each, or 10 weeks. This would increase a school's labor costs by just a few percent. The evaluation period would begin in the latter part of the semester. The evaluators could use the earlier part of the term to get the student profiles right and devise testing strategies -- which should not be known to the teachers.
This is just one potential solution to the educator incentive problem. There are infinite variations on this plan, but the point here should be clear. It is possible to determine a teacher's performance by using broader methods of evaluation and taking into account differences in student aptitude. It isn't easy to come up with a good solution, but it is possible. And working to devise a fair evaluation process is better than the alternative: an education system where subpar teachers are treated the same way as their superior counterparts.