The evidence behind strong claims about static vs. dynamic languages

As someone who intuitively feels that static typing must be helpful, I have to agree that the empirical evidence is not there - and I should not be surprised by that, as my own experience is fairly equally divided between the two paradigms, and I can't say that it makes a big difference.

On the other hand, in his analysis of E. Farrer's 'Unit testing isn’t enough. You need static typing too', the author says "The author does some analysis on how hard it would have been to find the bugs through testing, but only considers line coverage directed unit testing; the author comments that bugs might have have been caught by unit testing if they could be missed with 100% line coverage. This seems artificially weak – it’s generally well accepted that line coverage is a very weak notion of coverage and that testing merely to get high line coverage isn’t sufficient. In fact, it is generally considered insufficient to even test merely to get high path coverage, which is a much stronger notion of coverage than line coverage." OK, but is there empirical evidence that testing is actually practiced with anything approaching the rigor that he is setting as the standard of comparison here?