Thursday, April 8, 2010

Why Merit Pay is Worthless Concept In Education

Every once in while, politicians will pull the concept of merit pay out of a drawer, blow the dust off it, and hang it up in front of voters so that they can whack at it like a dime-store pinata. The fact that this particular pinata never seems to break open is probably a blessing in disguise, because I don't think anybody wants what is inside.





Taxpayers seem to think that merit-pay is a no-brainer, and that only poor teachers involved in scamming the system would protest it. After all, why would a good teacher be worried about getting extra money for their wonderful efforts in the classroom? The efforts to block merit-pay on the part of teacher's unions have routinely failed to make arguments that show teachers in a flattering light. Instead, we tend to come off as cynical time-servers who are willing to trade the minds of our future generations in return for job-security. It appears to many taxpayers that we are willing to protect incompetent teachers so that our own jobs will not be threatened.





In theory, there is nothing wrong with providing external motivation to inspire greater effort on the part of an employee. A worker in a factory, for example, may work just a little bit harder in order to earn a bonus due to their increased production. A salesman will hustle that much more for clients if they know that increased sales will result in healthy bonuses at the end of the year. Money talks, and it can be a powerful motivator. The problem is the act of teaching is fundamentally different from building a car, shoveling coal, or selling time-shares. The difference is not necessarily because teaching is more "noble" or "worthy" than selling vacuum-cleaners. It is simply because the definition of "success" for teachers is poorly defined, and the means by which students learn is complex and multi-dimensional.





In this respect, a teacher is similar to a doctor or a police officer. Let's say that we want to provide an extra financial reward for our best police officers in order to encourage a stronger police force in general. Fair enough, don't you think? The problem, howeer, is that we need to figure out a way to identify the "best" police officers from the merely competent ones and the merely competent ones from the incompetent slackers and time servers. After all, without some form of objective standard to judge effectiveness, we can't justify giving some police officers more money than others for doing the same job.



The easiest way to justify merit pay is to attach it to some numerical data that can be easily collected. For a police officer, it may seem logical to judge effectiveness by the number of arrests that are made by an officer. The more arrests you make, the more criminals you are removing from the streets and the more effective you are. The problem is that all arrests aren't created equal. Arresting an escaped murderer is quite a bit different from arresting someone for running a stop sign at an intersection. A police officer could inflate their arrest statistics by booking citizens for every little transgression of the law. In fact, it would make more sense in this case to concentrate on small-time offenses because such arrests could be made quickly and with minimum risk.



Perhaps, then, it would be better to judge police effectiveness by crime statistics for a geographic area. After all, if the police are effective at doing their job it makes sense that the overall crime rate in an area would decrease over time. This line of reasoning, however, is also flawed. The number of criminal acts observed in a community can rise and fall for all sorts of reasons. For example, the crime rate may drop because of large numbers of people start moving out of the area. Crime rates can rise because of economic conditions which cause people to lose their jobs, resulting in a greater temptation to make money through illegal acts. Crime rates can also be skewed by how police officers choose to make arrests. If crime statistics need to be lowered, police officers can simply stop making arrests for petty crimes that would normally be prosecuted.



None of this is a slam against the police. The same sorts of things can happen in lots of professions. If we tried to establish merit pay for doctors by rewarding those with the lowest number of patient deaths, we would create strong pressures for doctors to move away from treating the very sick. Right now in medicine we tend to reward those who perform complex procedures and surgeries, but the consequence is that less doctors choose to enter primary care where the financial rewards are lower. The unintended consequence is that more people cannot receive primary care, causing an increase in preventable illnesses and a decrease in the overall health of the population.



So now we get to teachers. Judging "merit" in teaching is extremely difficult to do in an objective, numerical fashion. Right now standardized test scores are the only possible way to get consistent, quantitative data about student learning. There are some major problems, however, with tying teacher pay to test scores:



First, the tests that we give to our students on a state level are terrible. They tend to be poorly written and few states have the monetary resources necessary to consistenty and regularly make changes to improve the accuracy of the scores. This means that the scores we get from these tests, even in the absence of cheating are not necessarily accurate.



Second, the way these tests are administered encourages cheating by schools. Schools whose very survival is based on test scores are given the tests well before they are administered to the students and told to give the tests without helping kids to cheat. The entire process is based on the honor system, and desperate schools can easily succumb to the temptation to "help" their students score better. In some cases this is the result of isolated cheating by individual teachers while in other cases it is a more coordinated effort at a school-wide level. Either way, the higher the stakes placed on the scores the more suspect the scores become.



The third problem comes up when you try to define success. Do you simply reward teachers whose students earn the highest scores? All this does is punish teachers who are working with our neediest students. It creates yet another strong motivation for teachers to avoid low-income areas with students who have the most to gain from the educational system. It rewards teachers who are working in wealthier school districts with students who already have many advantages coming in the door. Perhaps, then, you should reward improvement of test scores from year to year. The problem here is that one group of students cannot be easily compared to the group that comes in a year later. The variation in personalities, prior skills, home environments, and natural intelligence is far too great to simply look at improvement from one year to the next and draw useful conclusions. I have had years where nearly all of my students scored very well on standardized tests while the next year saw a huge drop in scores on the same test. This isn't because I stopped teaching, or because I put less effort and passion into my work. Often, the biggest variations in test scores from year to year occur because of natural statistical variations that I have no control over as a teacher.



So perhaps we should simply base "merit" on something other than test scores. Got any ideas?

No comments:

Post a Comment