"John Halpenny" wrote in message
oups.com...
One thing these figures seem to say is that 50% of SR20 accidents are
fatal, but only 10% of 172 accidents are.

As I pointed out previously, there aren't enough SR20 accidents (or even
SR20 and SR22 combined) to make any valid statistical conclusions. The
statistical error on the sample size exceeds the number of samples.

Pete