As a student pilot, often on my walk-around, I get the impression that there are any number of hazards that could pass me by, simply because I can look at every part on the checklist but I wouldn't necessarily recognise a problem with that part. This idea seems more plausible when I speak to more experienced pilots of the same type, and each of them warns me about checking one thing in particular that's not on the checklist, because they once saw an accident that was caused by a failure there - and crucially, each of them warns me of a different thing.
At the same time, when I read the monthly accident report bulletins, there are several every month where the cause was some part failing in flight, and they either imply that the failure should have been caught in a pre-flight inspection, or they explain why it might not have been visible. In one local incident a few months ago, a pilot made a hard landing and thought he made an extra-careful inspection afterwards, but made a subsequent flight without noticing that both blades of the prop had been shortened and the firewall was crumpled. More generally, taking off with your pitot cover still in place is the stereotypical example of a bad inspection.
Pre-flight inspections seem very error-prone, on the face of it. What evidence is there about how error-prone they are, and what fraction of defects they detect? I'm asking for either scientific experiments under controlled conditions, or statistical analyses of aircraft grounded after unsatisfactory inspections vs. failures in flight. Turning the question around, you might instead answer about how much less safe we would be in GA or air transport if we didn't have pre-flight inspections at all. This related question is pretty relevant, but focuses on checklist items which are forgotten completely: I'm hoping for a more general answer which also includes defects that are not detected even if the checklist is performed correctly.