14 trials is nothing - especially when even a 5% risk would be considered too high. For reference, the 95 % confidence interval on a distribution with a 1 in 5 accident risk would include this result:
https://epitools.ausvet.com.au/ciproportion
A 0.1% risk is probably considered too high for tourist flights that have no other useful purpose. They can't do enough flights to verify that experimentally, they'll calculate the risk via simulation and then validate the simulations with the test flights.
Indeed a 0.1% risk is far too high for tourist flights. The FAA doesn't accept the less than 0.0006% incident rate[1] of general aviation as safe enough for paying passengers.
