Another thing to consider. HP and torque can be impacted by the ECU's programming. This is something that CJV has noted a number of times. In short, when you do an exhaust comparo, unless you optimize (i.e., tune) the ECU for each exhaust system, you are leaving a lot on the table. Moreover, what's left on the table is somewhat random. The version of the GIAC program on Stephen's car, might, for example work pretty well right out of the box with the Fabspeed systems (or might have even originally been tuned with a Fabspeed in place). While the out of the box programming might not suit all of the other exhaust systems he tested as well.

One last thing... measurement error. I don't care how careful you are, there's going to be measurement error in dyno runs. With a 600 HP engine, 20 HP here or there translates into less than 3%... an error rate that I think is possible across dyno runs.. even possibly for the same car on the same dyno on the same day.