The current research explored whether lay people have a tendency to provide higher support for “wrong” predictions made by experts than those made by novices. Three empirical studies consistently revealed that there indeed exists a preference for wrong predictions even when predictions are made by experts. In addition, the current research also formulates preferences for wrong predictions made by experts in terms of a Bayesian inference and expresses the processes by which one may believe the wrong prediction in the form of two factors—prior odds and likelihood ratio. Finally, I argue that this preference is logical when treated as a result of the comparison between the two competing hypotheses.