I can't offer any studies either, but what I can tell you is what the organizations do who are dead serious about redundancy of connectivity.
1) Get multiple connections using different carriers
2) Look into having each connection use different technology (DSL vs. Cable, vs. T-1, vs. Satellite, vs. Wireless. Copper vs. Fiber, etc.)
3) Make sure that the circuits in question come into your facilities via different physical routes. In some cases, this has meant trenching to go under a different side of the property or installing poles to go over a different side)
The key point here is diversity in as many dimensions as you can get (or afford). At my home for example (because I have a home-based business), I have both DSL and Cable. While each has its own drawbacks and advantages, since the vendors, technology and media are different, the chances that both will fail simultaneously or for the same reasons are kept to a bare minimum. Similarly, even the firewalls are different (SonicWall vs. Cisco PIX).
Some instances based on personal experience (yes, I know that's not what you were hoping for) also indicate that since there is always some percentage of installers who are sloppy, or less than diligent, or worse, incompetent, that you perform (or hire) a complete physical audit of your networks looking for ANYTHING which is not as correctly done as possible. This includes loose terminations, equipment not on UPS (or at least good surge suppressors), equipment not properly mounted, any cabling not properly dressed (Eek! A naked cable), any cabling and equipment not properly labeled (type, vendor, purpose, etc.).
While there might well be such studies out there, my experience has been that careful planning and diligence in execution, and redundancy will pay far greater rewards than learning what the overall industry failure rates might be.
In your case, (as mentioned above) I'd see about finding out if there is any way of bringing in that second T-1 by a different physical route which avoids the manhole.
I know you wanted a study, not experiences, but I think what you might look for are documents on "Best Practices". Otherwise there are just too many variables out there.