Reference is made to the assumptions necessary in order to test a two‐state threshold model by an experiment using confidence ratings. It is argued that the testing of such a model by, aYES NOexperiment also involves corresponding assumptions. If these are taken a justified, then the fact that both types of experiment have been shown to give similar values of false‐alarm rate for a similar detection rate establishes that the appropriate assumption for an experiment with confidence ratings is justified. Consequently, either a two‐state threshold model cannot be tested by either type of experiment or it is legitimately tested by both.