Lies, Damn Lies, and Statistics
One of the respondents to a parallel thread on this site has cited a number of statistics and incident reports to demonstrate a point. We have previously cautioned against the validity of these arguments, as it is exceptionally difficult to place these incidents and causal factors in context. Statistics can be used to tell many different stories. As a very simplistic example, we provide the following:
The US AOPA Air Safety Forum site contains a collaborative link to the United States National Transportation Safety Board database, and enables viewers to access accident data tracking back to 1983.
The ASF data relates strictly to ACCIDENTS involving FIXED WING AIRCRAFT OF 12,500 POUNDS [5,700 KILOGRAMS] OR LESS in the United States of America.
A basic search of that database for the LAST TEN (10) YEARS found the following statistics:
Accidents where there were no injuries: 8818
Accidents where there were minor injuries (as the highest outcome): 2540
Accidents where there were serious injuries (as the highest outcome): 1724
Accidents involving fatalities (as the highest outcome): 3432
Total accidents: 16,514
This means that on average, there are 1,652 accidents per year in the United States, involving fixed wing aircraft of 5700 kilograms or less. On average, there are 343 fatal accidents involving fixed wing aircraft of 5700 kilograms or less per year.
A claim that has been made in your airspace design debate is that the United States represents best practice. This would seem to indicate that by the logic of your NAS architects, US accident rates of 1652 per year, and fatal accident rates of 343 per year are acceptable.
In (very) simplistic terms, if, as has been stated in documentation relating to your airspace debate, the amount of traffic in the United States is 20 times that of Australia, then using the logic of a proponent of airspace change in your country, an accident rate of 1652 divided by 20 – or 83 accidents per year - would be an acceptable safety target in Australia. That would translate to 17 being an acceptable target number of fatal accidents (involving light [GA and commuter] aircraft) per year.
Now, it is reasonable to assume that a fatal accident involves at least one fatality. So at least 17 deaths is acceptable. On reviewing statistics on fatal accident in the US over one recent month in last year (December 2003) it was found that out of 29 fatal accidents, there were 47 fatalities – so, on average, each fatal accident involved 1.6 fatalities.
This would seem to indicate, by the logic of your NAS architects, that 27 fatalities per year in light GA and commuter aircraft is an acceptable annual fatality rate in Australia.
This does not, of course, include statistics on accidents involving helicopters, and accidents involving aircraft at weights greater than 5700 kilograms.
Accessing the US NTSB site direct, it is possible to determine the total number of accidents that have occurred [as far back as the 1960’s] – including helicopters and aircraft heavier than 5700 kilograms. In 2003, there were 377 FATAL accidents in total in the United States – this includes all air operations, including Part 135 Air Taxi and Commuter Operations, and Part 121 Scheduled Operations.
Those 377 accidents resulted in 703 fatalities, at an average of almost 2 (1.9) per accident.
There were 2 fatal accidents involving Scheduled Operations, and 14 fatal accidents involving Air Taxi – Commuter Operations, for a total of 45 (fare paying passenger) fatalities – at an average of around 4 (4.2) per accident.
Using the “5% of the traffic” argument, and the suggestion that you should move to so-called best (United States) practice, it would appear that in Australia, you should be prepared to accept air accident fatalities at a rate of 35 DEATHS PER YEAR, with fare paying passenger fatalities at a rate of 2 DEATHS PER YEAR.
We must state that whilst the US ASF and NTSB statistics are accurate, the simplistic logic applied here in interpolating to Australia would NOT stand proper mathematical scrutiny – but then neither would the simplistic comparisons being done in other threads.
Given the amount of experience that the NTSB must have (over 1600 accidents per year), might it not be proper for them to examine your safety record, and operating conditions, carry out a proper comparative analysis, and suggest acceptable safety targets, or measure whether or not the safety levels apparently being achieved now in Australia are, perhaps, a better measure of “best practice”.