Outstanding Ron! You obviously have put a lot of heart and energy into this and have done a good job given the limited data available to you. Is there any way to find and associate flight time data?
It's difficult, especially considering the various sources of data.
The FAA General Aviation Survey says the typical homebuilt flies about 48 hours a year, while the typical 4-seat GA aircraft flies 94 hours. However, much of that GA aircraft time is for other than personal flight. For instance, the 2017 survey says personal flying in GA aircraft totaled about 7.8 million hours.... but the OTHER than personal-flight total is about 13.9 million. In comparison, the vast majority of EAB flights are personal...except in certain limited circumstances, they cannot be operated for hire or for business. Comparing flight hours is apples to oranges.
The FAA Survey results are based on the input from the survey, with the fleet size taken from the FAA registry. And hence, another problem. The process involves sending out a number of surveys to registered owners, and basing their calculations on the surveys returned.
Let's say they get responses from 60% of the EAB owners. A bit over half of the respondents say they are actively flying, with the remainder saying their plane is inactive. The FAA does the math, and decrees that, say, 58% of of EAB aircraft are active.
All well and good. Well, what about the 40 surveys sent out that DIDN'T get a response? They are ignored. So the FAA takes that 58% figure, multiplies it by the ~26,000 EAB aircraft listed in the FAA registry, and declares that 15,000 homebuilts are active. The survey says the average active EAB flies 48 hours a year, so the FAA computes that homebuilts flew 720,000 hours that year.
Not a bad process...when the FAA isn't making a separate effort to deregister inactive aircraft.
And how do they do that? By de-registering aircraft whose owners don't respond to a mailing almost identical to the survey.
Imagine, then, the FAA deregistering that 40% of survey targets who don't respond. All the computations are based on surveys that ARE returned, so the initial survey result is the same...58% of the EAB aircraft active, flying 48 hours per year.
But now, the FAA registry shows 22,000 EAB aircraft, vs. the 26,000 before because of all the deregistrations. The FAA AGAIN multiples the 22,000 by 58%, and says there are now 12,760 EAB aircraft. Multiply by 48 hours per year, and the EAB world flew 612,000 hours...a steep drop.
In reality, both cases had the SAME NUMBER OF ACTIVE AIRCRAFT, and about the same number of hours. The real-world inactive aircraft were (mostly) not included in the original survey, but the deregistration process produces an artificial drop in the fleet size, which the survey DOES utilize.
From 2010 to 2013, 23% of the US EAB fleet was removed from the registry. During the same period, the "active aircraft" percentage for the EAB aircraft in the FAA survey rose by just two percentage points. From 58% to 60%. The fact that ~7,000 inactive EAB aircraft were removed from the roster doesn't affect the results at all.
I personally got involved in this in 2013, at the peak of the deregistration program. The fatal accident rate for the EAB fleet had skyrocketed, and the FAA wanted to know why. It jumped because of the deregistration process...there were the same number of fatalities, but the OFFICIAL number of homebuilts had dropped by 23% due to the deregistration process. Note, AGAIN, the real-world number of active homebuilts hadn't really changed. But because of the combination of incompatible processes and a near-usual number of fatalities, it showed an artificial rise in the fatality rate.
Here's a plot that I showed the FAA at a meeting in 2013. Note, that, again, the FAA show more fatal EAB accidents because they count ultralights, SLSAs, ELSAs, an other aircraft as Experimental Amateur-Built.
So yes, I'm not too fired up about attempts to determine an accident rate per flight hour with the currently-available data. Computing the total flight hours for the fleet is fraught with the problems I mention above.
As I mention above, I do make a ROUGH CALCULATION on the yearly homebuilt utilization, and it does come in close to the FAA survey estimate (50.2 hours/year). Emphasis on rough...basically taking the aircraft total time when the accident occurred (when the NTSB report includes it), the date of the accident, and the "Model Year" of the aircraft.
That should make LolPilot's statistics students cringe. Just the fact that the model year is an integer throws the results around wildly.
In my defense, I don't use it in any external calculation. My assumption is that the integer model year issue balances out... that there are probably about as many airplanes completed EARLY in the model year, as the late. I basically use it to compare the approximate yearly utilization rate for different homebuilt models.
Finally, I do incorporate the median pilot total flight hours into my analysis. It does make a significant impact. For instance, the rate of what I call Pilot Miscontrol (stick-and-rudder errors by the pilot, vs. errors in judgement) is lower than the Cessna 172...but the median total flight time for pilots involved in homebuilt accidents is more than twice as high.
Ron Wanttaja