Evaluations that use imperfect information run into both analytical and political problems. Educational accountability systems based on achievement scores are an instance. Such systems frequently turn out to be irrelevant to policy decisions, resisted by educational interest groups that fear unflattering comparisons and the misuse of results, and…