Network tomography tools such as MINC can be used to infer important internal properties of a network underlying a multicast tree. This inference is made by analyzing receiver feedbacks to the measurement probes sent from the source and can be utilized to make significant decisions concerning the network. However, certain misbehaving receivers can return incorrect feedback and mislead the MINC inference resulting in an erroneous decision. Hence it is required to verify if the feedbacks collected from the receivers can be utilized to make a trustworthy MINC inference. In this report, the MINC inference procedure which computes the loss probabilities of various paths in the multicast tree is investigated. Firstly, it is shown how the loss probabilities computed by MINC change when receivers alter their feedback. Then, a statistical detection procedure is presented, which searches for loss probability inconsistencies in the feedback data. Some preliminary results b! ased on old mbone loss traces are presented.