Well, in isolation p is problematic. For instance, without hypothesis pre-registration you have the obvious problem with negative results and publication bias.