simulators don't replicate the 'pilot experience'
This is the main problem with sims ... how close is the box to reality (ie what is the fidelity). In the final analysis the only "true" simulator is the aeroplane .. all the ground based boxes are in error to a greater or lesser extent. For routine airline use, the simulator fidelity has to be "good enough" for the purpose but the reality is that, outside the operating regions tweaked for the purpose, the box might be VERY far from reality.
Most pilots never get a chance to participate in REAL performance limiting takeoff and landing trials .. I don't use my descriptors lightly .. first time I was involved (with an experienced OEM experimental TP driving) I nearly embarrassed myself ..... good fun but VERY different to normal line landings .. which is why I often suggest that pilots ought to respect the 1.67 factor for normal landings ..
Keep in mind that
(a) the test folk don't just head out to run some trials today because they had nothing much better to do .. the crew will have done a moderate series of practice runs over the previous days to make sure that the polish is well buffed
(b) on the day, the TP will nominate a number of trials to be discarded for whatever reason ...
(c) when we get to run the post trials analyses, we engineering folk toss out a few more which don't suit the purpose ..
(d) end result is the data presented is skewed somewhat toward the less conservative .. one of the reasons that the AFM fudge factors are required.