We study the role of beliefs about experts' honesty in a market for credence goods with second opinions and overtreatment. Experts are honest or dishonest. The population shares a common belief about the share of honest experts, which may be incorrect. We characterize the belief that maximizes consumer's expected utility and show that it is generically different from the true share of honest experts and larger than the one that maximizes the equilibrium level of honesty. We then analyze the decision of an authority that has learned the actual share of honest experts whether to publicly reveal it through a costless information campaign, thus correcting people's beliefs, and show that it does not depend on how wrong beliefs are. We further show how increasing market transparency (making experts more aware of the number of opinions collected) affects the optimal belief and may have a positive as well as a negative effect on overtreament. Finally, we briefly see how a successful campaign run in Switzerland in the mid '80s to reduce excessive hysterectomy rates can be read through the lenses of our model and how accounting for beliefs about honesty might allow theoretical predictions to better fit experimental evidence.