Performance-related pay

I have reading about the Medicare/Medicaid demonstration project which has started to reimburse doctors based in part on the “quality” of their care. To incentivize better results, the theory goes, doctors whose patients’ health improved could get reimbursed at a higher rate. Ah, but how to measure quality fairly, I hear you ask…what about all the things that doctors can’t control? Patients who are obese, patients who don’t even bother to fill their prescriptions… Surely a doctor can’t be blamed if these patients fail to thrive? In education, approaches to increasing the precision of performance estimates include using test scores from multiple years of classes (in other words, increasing the sample size for the estimate); combining value-added scores with other, independent measures of teacher performance, such as Headteachers’ evaluations; and calculating scores at a higher level of aggregation (e.g., for all the teachers in a given subject or for all the teachers in a school, which, again, increases sample size). The model would have to be very complex; few teachers would buy into it and, those that did, would probably find a way to game the system. Many resent the intrusion. Some may drop low-performing students. Others may indeed respond to the incentives—for ego or financial reasons—and actually improve their practice to lead to better outcomes. I welcome the ideas but how it will work in practice. So far, there’s not a lot of evidence that using value-added models to evaluate teaching actually improves outcomes for kids in the real world. There’s not a lot of evidence that it doesn’t. Time may tell.