I'd love to know as well. As far as I know (and I happily stand to be corrected) you lose mechanical advantage the higher the gear. On any good dyno, runs done one after the other in progressively higher gears show almost parallel lines that get lower for each higher gear.
I'd actually like to tack on a question here since you're all in the tuning game. My understanding (which is probably wrong) is that the car, on the street, actually puts down the same amount of power regardless of gear, but a dyno measures power by measuring the rate of change of speed comared to how quickly the car climbs to X RPM, and punches the math behind it. There is no real way to measure actual torque as it is on the surface of the tires, just mathematical induction. As soon as you, say, reduce the weight of the rollers, the values would rise. It is obviously not possible for a car to climb as quickly through 5th as it can 4th, 3rd, whatever, which is why these rolling dyno's release lower numbers for higher gears, because of the rate at which the car climbed through the RPM range.
Dyno's are purely used for comparison before and after tuning. As far as real correct figures are concerned, I don't think that dyno's are ever 100% accurate. They're used as a means of comparison on the same car, on the same day, in the same gear, with different tunes/setups. When you start comparing it to other cars, even if they're the same make/type/year, you're missing the point. Same goes for choosing a different gear. You should pick the gear closest to 1:1 to avoid torque multiplication in the gearbox, or it colours your results.
AmIcorrectorwhat?
...
