Teachers, Cheating, and Incentives
In recent years there seems to have been a surge in academic dishonesty in high schools. No doubt this can be explained in part by 1) increased vigilance and reporting, 2) greater pressure on students to succeed, and 3) the communicable nature of dishonest behavior (when people see others do something, whether it’s enhancement of a resume or parking illegally, they’re more likely to do the same). But, I also think that a fourth, and significant, cause in this worrisome trend has to do with the way we measure and reward teachers.
To think about the effects of these measurements, let’s first think about corporate America, where measurement of performance has a much longer history. Recently I met with one of the CEOs I most respect, and he told me a story about when he himself mismanaged the incentives for his employees, by over-measurement. A few years earlier he had tried to create a specific performance evaluation matrix for each of his top employees, and he asked them to focus on optimizing that particular measure; for some it was selection of algorithms, for others it was return on investment for advertising, and so on. He also changed their compensation structure so that 10 percent of their bonus depended on their performance relative to that measure.
What he quickly found was that his top employees did not focus 10 percent of their time and efforts on maximizing that measure, they gave almost all of their attention to it. This was not such good news, because they began to do anything that would improve their performance on that measure even by a tiny bit—even if they caused problems with other employees’ work in the process. Ultimately they were consumed with maximizing what they knew they would be measured on, regardless of the fact that this was only part of their overall responsibility. This kind of behavior falls in line with the phrase “you are what you measure,” which is the idea that once we measure something, we make it salient and motivational. In these situations people start over-focusing on the measurable thing and neglect other aspects of their job or life.
So how does this story of mis-measurements in corporate America relate to teaching? I suspect that any teachers reading this see the parallels. The mission of teaching, and its evaluation, is incredibly intricate and complex. In addition to being able to read, write, and do some math and science, we want students to be knowledgeable, broad-minded, creative, lifelong learners, etc etc etc. On top of that, we can all readily agree that education is a long-term process that sometimes takes many years to come to fruition. With all of the complexity and difficulty of figuring out what makes good teaching, it is also incredibly difficult to accurately and comprehensively evaluate how well teachers are doing.
Now, imagine that in this very complex system we introduce a measurement of just one, relatively simple, criteria: the success of their students on standardized tests. And say, on top of that, we make this particular measurement the focal point of evaluation and compensation. Under such conditions we should expect teachers to over-emphasize the activity that is being measured and neglect other aspects of teaching, and we have evidence from the No Child Left Behind program that this has been the case. For example, we find that teachers teach to the test, which improves the results for that test but allows other areas of education and instruction (that is, those areas not represented on the tests) to fall by the wayside.
And how is this related to dishonesty in the school system? I don’t think that teachers are cheating this way (by themselves changing answers, or by allowing students to cheat) simply to increase their salaries. After all, if they were truly performing a cost-benefit analysis, they would probably choose another profession—one where the returns for cheating were much higher. But having this single measure for performance placed so saliently in front of them, and knowing it’s just as important for their school and their students as it is for their own reputation and career, most likely motivates some teachers to look the other way when they have a chance to artificially improve those numbers.
So what do we do? The notion that we take something as broad as education and reduce it to a simple measurement, and then base teacher pay primarily on it, has a lot of negative consequences. And, sadly, I suspect that fudging test scores is relatively minor compared with the damage that this emphasis on tests scores has had on the educational system as a whole.
Interestingly, the outrage over teachers cheating seems to be much greater than the outrage over the damage of mis-measurement in the educational system and over the No Child Left Behind Act more generally. So maybe there is some good news in all of this: Perhaps we now have a reason to rethink our reliance on these inaccurate and distracting measurements, and stop paying teachers for their students’ performance. Maybe it is time to think more carefully about how we want to educate in the first place, and stop worrying so much about tests.
(This post also appeared as part of a leadership roundtable on the right way to approach teacher incentives in the Washington Post. The Washington Post will post more opinions about this topic here. )