Here's the paper that Robert Ward of Maryland shared with me via email last year. The stats look solid to me, although I didn't check his p-values. Let me know if you work this into any of your projects this fall!
Now that the NSA has declassified this information about cryptanalysis, either they can no longer utilize these methods of cryptography, or must do so with caution, knowing that anyone has the resources to crack them. Was it ethical of the NSA to release this information, especially if other countries still use these methods to create ciphers?