Broadcast schemes play an important role in the performance of mobile ad hoc networks, which are a clear example of ubiquitous wireless multi-hop networks where nodes collaborate in a distributed way. They are widely used as a dissemination mechanism and as a part of the discovery phase of routing protocols. The simple flooding algorithm is the usual mechanism employed in mobile ad hoc networks, but its inefficiency has been demonstrated in congested scenarios due to the high number of collisions and contentions. However, these problems can be partially alleviated by using a probabilistic broadcast approach in which every node forwards the incoming packets according to a certain forwarding probability. In this paper, we use a simple probabilistic broadcast protocol to evaluate the effects of congestion on the performance of broadcasting in ad hoc networks through a mediation analysis. We hypothesize that the congestion mediates in the relationship between the forwarding probability (independent variable) and the output metric (dependent variable). We consider several output metrics according to the application of the broadcasting protocol such as reachability, broadcasting delay, packet delivery fraction and end to end delay. The simulation results show the existence of the mediating effects and how such effects may be counterbalanced depending on the target use of the probabilistic broadcast scheme.