In centralized public school admissions, participants may not fully observe other participants' preference rankings and priorities. In this paper, we evaluate prominent assignment rules in terms of their auditability properties. We show that Serial Dictatorship and Immediate Acceptance rules are maximally auditable, in a sense that any deviation from running them as promised can be detected with only two applicants' private information. The Deferred Acceptance and Top Trading Cycles rules are minimally auditable, in a sense that some deviations will not be detected unless one has close to full information about the preference rankings and priorities.