The news that an Atlanta schools chief spearheaded a vast criminal conspiracy among administrators and teachers to cheat on standardized tests to artificially boost the scores of their pupils is as depressing as it is unsurprising. And it reinforces my belief that merit pay systems for teachers that are rigidly tied to student test scores are a terrible idea, even if school districts are using very sophisticated value-added analysis. What we should be pressing for is more business model innovation, in which different charter networks and schools and specialized course-level instructional providers are allowed to pursue distinctive compensation strategies. This wouldn’t eliminate cheating, as charter or portfolio districts of the kind I have in mind would provide parents with test scores, and parents would reward schools that appear to deliver better outcomes. But this problem is easily addressed: because innovative schools would have a higher hurdle to clear than traditional schools in terms of winning parental trust, they should be required to use third-party proctors to administer tests. It’s kind of absurd that this practice isn’t more widespread.
My guess is that the scandal in Atlanta will lead some teachers union allies (a) to condemn testing as such, (b) to press for higher compensation for all teachers union members, irrespective of student outcomes, and (c) to insist that the only way to improve student outcomes is to dramatically increase spending on social services, ideally social services delivered by public employees. Though there might be a case for somewhat higher spending on social services in some places and in some cases, a more intuitive view is that we should make it harder for administrators to cheat.