Yes, HammerTime, we do know that the response of all the aluminum stems over all reasonable loads is linear. This is a basic tenet of Hooke’s Law as well as the mechanics of materials class typically taken by sophomore-level engineering students. The linear assumption is valid as long as you’re talking about homogeneous, isotropic, linear elastic (HILE) materials under small strains. These assumptions would be invalid if the stems were yielding (i.e., not bending back to the original shape once the load is released) or if the bending deflections were large (think fishing rod bent roughly into a parabola). There’s more information here:
Technically, carbon composites are neither homogeneous nor isotropic. However, the structural response of composite structures under small strains is typically linear. Given the testing protocol, these results are valid for both aluminum and carbon stems.
You ask if some stems might be more linear than others. Well, yeah, kinda. A carbon stem may well have one stress/strain curve as it’s loaded and another stress/strain curve as it’s unloaded. These two curves have different areas beneath them. The area under the curve represents energy. Since the loading curve has more area beneath it than the unloading curve, not all of the energy from the deformation is being returned during unloading. The technical term for this phenomenon is “hysteresis,” and many riders perceive hysteresis as improved vibration damping.
But this test involved only loading; no measurements were taken on unloading. So the linear assumption is valid for both aluminum and carbon stems.
Edit: fixed a sign error and improved clarity.