I'm pretty certain that most managers know what they are talking about most of the time, but, despite applying tools, techniques and frameworks diligently, still find themselves in trouble with their projects.
I happen to think that a big driver of project problems is not that tools are inadequate but that deep skill is needed to select the right ones for the circumstances. Methodologies and standard processes (such as PRINCE) assist by pre-selecting a range of tools, but success here depends on the conditions in the project at hand being consistent with the assumptions in the methodology.
Assuming that managers can compentently select appropriate control paradigms, awareness of the project environment and internal health is critical. Risk Awareness is one case in point, one that I think is undervalued at present, despite the apparent depth of numerical frameworks. I wrote about this before, showing my framework for the descision space of all projects.
Here's another project classification I like, since it focuses on the nature of uncertainties that may be met in a spectrum of projects.