The eternal frustration for anyone who has studied general aviation safety statistics is the complete lack of reliable data on exposure. We know with reasonable accuracy how many accidents occur each year, and we know with even better certainty how many of those wrecks cause a fatality. But how many hours were flown, how many airplanes were active, how many takeoffs and landings, or any other activity data for the general aviation fleet is a guess at best.
The accident rates – number of accidents per 100,000 hours of flying – that you see in reports from the FAA and NTSB are based on an annual general aviation activity survey the FAA conducts. The survey is now conducted primarily online instead of through the mail and it asks a selected minority of registered airplane owners how much they flew in the last year, under IFR, night, number of landings, and so on. The survey asks for answers to all of the questions I have, but I don’t believe in the results.
After owning an airplane continuously for at least 25 years and never having been selected to participate in the activity survey, I have been selected each year for the past 10 years. I don’t know why that happened. Maybe because I have completed the survey each time I was asked that puts me into some sort of favored list of reliable responders. But if I’m answering the survey every year, and my patterns of flying vary little, that means other airplane owner-pilots are being left out.
I think the biggest problem with the survey is that it asks for “facts” that are part of an emotionally driven topic – how much you fly. Ask a person how much they drive each year and you get a pretty unemotional answer because nobody finds extra status in driving more or less. But ask a pilot how much he flies and you are way into status territory. You might as well ask us how much we weigh, or little we drink or smoke, or how big our, well, you-know-what is. How big is our hangar? What do you think I meant?
The FAA in its survey tries to eliminate the emotional component by instructing us to collect up log books and do the math before answering. But I confess, I don’t do that. I just wing it. I come close, I think, but I bet if I do err it’s on the side of flying more than I really do.
When all of those survey errors line up in the same direction the results are a number of hours flown that is inflated by who knows how much. I once did the math of how many total hours pilots reported they had flown and compared that to the number of hours the survey found the fleet of airplanes had flown. All I can say is that there are many pilots flying hours in something other than airplanes because the pilot hours totaled to a much larger number than the reported hours flown by airplanes.
When it comes to the airlines the FAA and NTSB have very accurate records on the number of hours and departures flown so accident rates for the airlines are precise and reliable. Even in automotive safety data the results are reasonably reliable because the sample is so huge and exposure data is collected by many sources including insurance companies, state regulators, and manufacturers. But in general aviation we just don’t know with reasonable certainty how many accidents happen per unit of exposure.
A couple years ago the FAA acknowledged how unreliable its accident rate data is – particularly for experimental aircraft – and decided to change its focus to reducing the total number of accidents no matter what exposure may be. That makes perfect sense because information on how much experimental airplanes fly is the least reliable of all, but the count of fatal accidents is as accurate for experimentals as for any other category of airplanes.
Since 1989 the number of fatal accidents for all of general aviation has been in steady decline from 487 in 1989 to 268 last year. I believe there have been some improvements in GA safety over those years, but there has also been a very steep drop in flying activity. But nobody knows for sure how much activity dropped, but we do know there are almost half as many fatal accidents in all of GA compared to 20 years ago.
It is a very different picture for the experimental airplane fatal accident count. There were 57 fatal accidents of all types of experimental aircraft in 1989 but the total hit 88 in 2009. The fatal accidents were down to 65 last year, but for the 20-year period the number of fatal accidents involving experimentals has grown at least slightly while the total for all of GA has dropped significantly.
In an attempt to be fair to those flying experimental aircraft the FAA decided to set a metric – a numerical goal – based on fatal accidents without an attempt to guess at an accident rate. The baseline number of fatal crashes per year for all experimentals for the past five years is 73. To meet the FAA’s overall goal of a 5 percent annual improvement in safety the “not to exceed” fatal accident number for fatal accidents in experimentals for this year is 70.
The “year” is the federal government’s fiscal year that begins on October 1, so we have only a couple of weeks to go to the end of this year. The good news is that fatal experimental accidents are below the not to exceed total of 70, but just barely. I don’t have the latest count but a couple more accidents will put us over the goal.
Using all experimental aircraft to establish the safety metric is a bit unfair to amateur-built airplanes because there are accidents involving experimental exhibition airplanes and developmental airplanes from manufacturers. But the huge majority of accidents in the overall category are AB so that is where the risk lies and improvement must come.
What happens if the fatal accident count by the end of September exceeds 70? Nothing specific, but you can be sure there will be more scrutiny of experimental flying and the potential for new regulation.
I think the safety metric based on accident count is reasonable and is at least definable, something missing in the past. While it would be great to eliminate accidents, the realistic goal is to show improvement. If the accident total trends the wrong way, it will be very difficult to defend.