Originally Posted by
Sunfish
As for the two AF fatalities, if they were statistically significant ATSB would have howled it from the rooftops.
You keep saying that, but they did find it was statistically significant. There are about 40 pages of analysis in the report. For better or worse, someone picked out one easily understandable number to put in the media release rather than the 40 pages of analysis.
They didn't just analyze 2 flights, they compared 13,389 angel flights with 2 fatal accidents and 3,527,079 private flights with 72 fatal accidents. If you want to argue statistical significance the numbers are there - do the maths or consult a statistician and tell us what you come up with.
If you want to argue statistical significance, you also need to decide what confidence you are looking for. Statistical significance is a measure of how likely the result is to have occurred by chance, versus having occurred due to a real difference.
If you are asking whether Angel Flight need to improve safety, you might settle for a lower level of confidence than if you e.g. were introducing a new medical treatment.
How sure would you need to be that Angel Flight has a higher accident rate before you tried to improve safety?
80% sure? 90%? 95%? 95% is a common level to be considered significant but it's probably low if you are evaluating medical treatment and probably too high if you are worried about a safety problem.