There isn't much. How many companies are going to publish papers of the form
"Here's how we failed to improve our process having blown two years effort"?

Try the SEI's Information Repository at
http://www.sei.cmu.edu/seir/index.html

"The Software Engineering Information Repository (SEIR) is designed to serve the software engineering community in the role of gathering, coordinating, analyzing and disseminating, data and information on the impact of software engineering practices as well as practices and innovations leading to improvement. " - SEI.

The public area of the actual repository is at http://seir.sei.cmu.edu The interesting stuff is in the members only area (isn't it always), but last time I looked registration was free of charge.

Just a thought: I think much of the data on CMM/ISO/Etc success/failure rates is noise. All these schemes are bodies of knowledge capturing how people define and improve processes. Their success or failure lies in how people interpret these complex bodies of knowledge, not the bodies of knowledge themselves. I'd bet most failures are due to politics, most successes due to a succesful leap of faith and excellent team management skills, but since I'd never be able to prove it, I won't. The success rates reflect humanity's ability to express complex thoughts clearly.