Safety Statistics

[framer]

On the Turkish accident thread someone posted a link to some stats on fatality events for different airlines. The stats were from 2004. The stats did not include hijackings. The figures are the number of accidents involving at least one fatality, that occur per million flights.If an airline had not had a fatal event it was not listed.
Of the 60 airlines, the best/safest ten airlines were;
1/ All Nippon Airlines 0.22
2/Delta 0.30
3/ British Airways 0.32
4/ Lufthansia 0.41
5 NorthWest Airlines 0.43
6/ Aloha Airlines 0.49
7/ United Airlines 0.50
8/ Mexican Airlines 0.53
9/ US Airways 0.56
10/ American Airlines 0.59

The ten worst rated airlines were;

1/ Cubana 24.0
2/Aero Peru 16.7
3/Air Zimbabwe 12.5
4/ China Airlines 10.2
5/ Royal Jordanian 8.82
6/ Egypt Air 8.00
7/ Turkish Airlines 7.30
8/ Air India 6.82
9/ Value Jet 5.88
10/Korean Air 5.38

My question is; Is there any group or safety committee dedicated to looking at statistics like these and determining what the good ones have in common and what the poor ones have in common?

PS. Lies, damned lies, and statistics (just thought I'd save someone the trouble)

[kenparry]

My question is; Is there any group or safety committee dedicated to looking at statistics like these and determining what the good ones have in common and what the poor ones have in common?

Yes. Flight Safety Foundation does a lot of work in this area. A decade or so ago they started a mentoring programme, trying to team up weak carriers with safe ones, to encourage a safety culture. Not sure how it is going, as I lost track when I retired from airline flying.

Try this for more info:

Welcome to Flight Safety Foundation (http://www.flightsafety.org/home.html)

[PEI_3721]

The following report looks a fatal accidents, it breaks out world locations and aircraft types; the originators of course know who the operators are.
The other aspects of the analysis, such as the causal factors could be of interest, particularly if cross referenced to an operator, e.g. ‘flight crew – omission of action/inappropriate action’, for each airline.
Global Fatal Accident Review. (http://www.caa.co.uk/docs/33/CAP776.pdf)

[Rainboe]

On the Turkish accident thread someone posted a link to some stats on fatality events for different airlines. The stats were from 2004. The stats did not include hijackings. The figures are the number of accidents involving at least one fatality, that occur per million flights.

Have I missed something? What fatalities has BA had since 2004? Are they including 'death from natural causes' inflight mortality in these figures, otherwise the statistics appear to be a little dodgy to me!

[framer]

Sorry Rainbow,
I can see how it reads like that. My mistake.
The stats were released in 2004 and are based on the total number of flights the airline had carried out up to 2004.
eg if the airline had carried out two million flights by 2004 and had had one accident where a fatality was recorded then the number beside the airline would be 0.50.
Does that make more sense?
Regards, Framer

[framer]

Are they including 'death from natural causes' inflight mortality in these figures
No.

There is a massive amount of info available to us nowdays isn't there.

I would like statisticians to run equations like;
If your airline is listed in the top ten then the probability

1/ That you are operating in US airspace is.........
2/ That your pilots earn above average pay is..........
3/ That your CEO is from Tahiti is...........
4/ That you have less than 3 a/c types in the fleet is........
5/ That you operate FOQUA is.........
6/ That your 6 month CRM training is two days long is.......
7/ That you have an established fatigue monitoring system is.......
8/ That you operate with mixed first languages in the flight deck is.....
9/ etc etc etc

Maybe it's been done a thousand times I don't know. It just struck me that an international group should use the math to come up with a list of recomendations that companies can aim for to make their operations safer.
Form a group whose only job is to create such a list and then they disband. That way it won't be lost in yet another organisation. Take representatives from FAA, CAA, NTSB, etc etc and give them access to some wizzo mathmaticians. Tell them their only goal is to produce a document identifying the top twenty things (statistically) that an airline can do to become one of the safest. That way airlines can identify their threats, ie if for example operating in US airspace was one of the factors and the airline couldn't do that, they could then think about ways to mitigate the increased risks presented by their own airspace......training, awareness, etc etc.

[PEI_3721]

framer, these are all good points in principle, but in practice a deep statistical review may not result in an improvement. 'Name and Shame' might not work in those areas where poor safety results for poor corporate or national authority capability – they don’t care.
Similarly, there might be biases due to using fatalities as a reference. The current statistical method might show an airline as having a good safety record, but a review of serious incidents could indicate otherwise. Consider if an unfortunate passenger drowned in the Hudson vs a small change in the parameters at AMS enabling all to survive, who then has the better record.
Statistics are a good tool, but in encouraging change to improve safety they must be used with other tools, but I hasten to add that I wouldn’t know what to recommend.