Perhaps the successful entrepreneurs share another interesting trait. This trait is detected by good mentors, which leads to a correlation between entrepreneurs being successful and having good mentors.

the statistical term is sample selection bias. to test the importance of mentors you'd need to figure out some way to compare to groups of start-ups which are identical in every way except access to mentors... which i can't think of a good way to do... maybe if you could find start-ups that had been accepted by the same vcs.. so maybe one vc gives more money the other vc had more 'mentor's then you might be able to reasonable estimate the effect of vcs... but even this isn't great.

I was referring to more than just the mentor. Such as the ideas, practices and culture of their peers.