Potty Page

August 21, 2008

Incorrect use of statistics?

I've just been reading news articles about the crash of a Spanair MD-82 in Madrid. Loads of news sources (I'll pick this one suggest that the plane is currently the second safest - this statistic comes from airdisaster.com.

I went to check to see if the Boeing 777 was down as the safest - with the prior knowledge that it has had no crew or passenger fatalities to date. I was surprised to see that's it's not... the award for safest plane goes to the Saab 340 (which has had 0.333 fatal incidents per million flights). The 777 isn't listed, which suggests that planes that have had no fatalities are not considered in the safe list... so in reality the MD-82 is considered the 2nd safest plane of all those that have had killed people, which isn't as greater claim to fame.

I don't think the Airbus A380 has killed any passengers or crew (I keep wanting to type cast and crew...) yet either...

Missing the data of when things go ok reminded me of a paper, Risk Analysis of the Space Shuttle: Pre-Challenger Prediction of Failure. Where it appears that when trying to work out if frozen O-rings are bad or not it was neglected to take into account the temperature of launches when there was no deterioration of the O-rings. If you ignore the data when the were no problems, then the failures look fairly unrelated to temperature. However, the lowest temperature they'd launched without having any deterioration was 66 degrees Fahrenheit, below 66 they'd had always had some deterioration - once they launched at 53 degrees (STS-51-C) and they had 3 failures... Challenger was launched when it was 31 degrees.

Posted by Ed at August 21, 2008 5:27 PM | Rant |