Testing methods could have something to do with the test results. I could see overly high data rates with compressible data, assuming that data compression is used not only on the EV-DO interface, but the T-1's as well. I do agree that the results do not agree with the physical device limits, and should not be considered legit.
