By Ryan McNeely
Many people are wary of the increased reliance on standardized testing to measure student performance, a key component of No Child Left Behind and the education reform movement. While it is useful to have a universal standard to measure how well students are learning relative to one another, yesterday’s New York Times contains an account of a perverse incentive created by high-stakes testing: teacher and administrator cheating. Both the principal and assistant principal of a Houston elementary school were forced to resign after it was discovered that they had conspired to give their students an advantage. The scheme wasn’t particularly ingenious: they simply bent the sealed test booklets far enough so they could see the questions, and then created study guides based on those questions. The students’ scores were invalidated.
But this isn’t the first time teachers have been caught manipulating test results. Brian Jacob and Steven Levitt released a study in 2003 showing that when teachers have an incentive to cheat, some will do so. Invited into the Chicago public schools for a three-year investigation by then-Chief Education Officer Arne Duncan, Jacob and Levitt used a statistical algorithm to demonstrate that between “3 to 6 percent of classrooms experience instances of teachers or administrators’ doctoring students’ exams.” Note that the study only examined instances of very crude cheating—teachers literally changing answers on scoresheets—and did not account for other less egregious forms of cheating like the Houston “study guides” or simply giving students extra time on exams.
One response is to simply minimize the problem. For example, Beverly L. Hall, superintendent of the Atlanta Public Schools, responded to the Houston scandal by saying that instances of cheating are rare, and that “teachers over all are principled people in terms of wanting to be sure what they teach is what students are learning.” A smarter response is to simply acknowledge that while assessing and acknowledging difference in teacher quality is both promising and important its promise will be undermined unless proponents address cheating. The good news is that the Jacob and Levitt study demonstrated that even very simple oversight measures, such as making teachers aware that scores will be scrutinized for irregularities, can dramatically reduce cheating frequency. As the trend toward greater reliance on tests increases, responsible policymakers will have to remain vigilant about the potential for cheating and inform themselves about the most effective countermeasures.