Theories of political accountability assume citizens use information about the performance of government to hold public officials accountable, but whether citizens actually use information is difficult to directly examine. We take advantage of the importance of citizen-driven, performance-based accountability for education policy in Tennessee to conduct a survey experiment that identifies the effect of new information, mistaken beliefs and differing considerations on the evaluation of public officials and policy reforms using 1,500 Tennesseans. Despite an emphasis on reporting outcomes for school accountability policies in the state, mistaken beliefs are prevalent and produce overly optimistic assessments of the institutions responsible for statewide education policy. Moreover, individuals update their assessments of these institutions in an unbiased way when provided with objective performance data about overall student performance. Providing additional information about race-related performance differences does not alter this relationship, however. Finally, support for specific policies that are intended to improve student performance is unchanged by either type of performance information; opinions about policy reforms are instead most related to race and existing partisan commitments.