Thank you very much for doing this experiment and for sharing the results.
Regarding this part:
That is only valid with some strong assumptions: independence of the observations, low variance and some sort of stationarity (i.e. that results do not depend on time and location). I am not sure...