Eboy
The statistics you linked to are flawed and tend to exaggerate the values for airlines with a greater volume of flights. It may be that this was done unintentionally, or it may be that it was done so as to give some spread in those airlines who have had no recorded fatalities. Either way, it is not an appropriate measurement for comparison. In actual fact, there is no appropriate measure for comparison of individual airlines because the probability of a "fatal" flight is so small compared to the number of flights that most carriers perform.
The figures used are simply the difference between the expected number of fatal flights and the actual adjusted fatal flights. A brief dimensional analysis shows that this figure has "dimension" of "adjusted fatal flights" (for want of a better description) which is not independent of total flights for the airline and therefore does not give a fair comparison.
A better way of comparing would have been to simply divide the adjusted number of fatal flights by the number of flights. This yields a number which is equivalent to the probability of a fatality for a single passenger per million flights.
If you use this method, although it still shows all US carriers together have a net fatal accident rate (0.33) lower than European (0.37), the difference is not statistically significant, and can easily be attributed to random factors.
Interestingly, if you assume a normal distribution for fatal flights among airlines, there are only two airlines that fall outside the 95% probability for the figures being just as a result of random factors: China Airlines (marginal at 2.32 standard deviations from mean) and Cubana (weighing in at a massive 8.23 standard deviations from mean).