I have a theory about power production and recording on bicycles. Maybe I should call it an observation. At higher speeds like on a TT bike, there is more wind resistance relative to rolling resistance. Higher wind resistance takes away the bit of "swing" in your pedal stroke - the very, very slight surge in each stroke. I think power meters report this as lower power. If you are somehow able, see what your TT bike power is up a 20, 30 or 60 min 4% mountain grade. I would predict it's higher. I find my power is highest on a cyclocross type bike with fat 35mm gravel tires. Next a regular road bike. And last my triathlon bike. This is also why on most home mag trainers, it's impossible to match outdoor power. The mag trainer has a unnatural 360 degree resistance that takes away much of that slight surge with each rotation of the crank.
There's a difference between what are errors in measurement and what are differences in physiological capability under varying environmental and riding scenarios.
The only reason a lower inertial load may erroneously inflate power meter reported data is if it results in a significant enough violation of the assumption of constant crank rotational velocity used by crank/pedal power meters to calculate power output.
Assuming circular chainrings, at most
the error in power will be of the order of 2-3%, and that's for the extreme scenario of the initial couple of pedal strokes of a maximal acceleration from a standing start. Steep climbs and low inertia trainers are the other scenarios where the crank velocity does vary more than flat road riding but the potential power meter error resulting form these are less than half of the acceleration scenario.
The crank velocity variations in all other (circular chainring) scenarios will be even less, and the resulting power meter error will be negligible.
Here's some detailed assessment of that issue if you are interested in reading more about it: http://alex-cycle.blogspot.com.au/...-crank-velocity.html http://alex-cycle.blogspot.com.au/...-crank-velocity.html