Feedback on 10 year FAA SLSA safety report

[Merlinspop]

Bruce, you are correct. My over 50% should have said 50% or about 50%.

I don't have a dog in this fight. I love the look of low wing planes, and it may be what I buy. But I want my eyes wide open, and I think the data is too clear not to pay attention.

Yup. Goes back to my earlier comment. Do your own research, talk to people as you're doing here (but don't sweat it if there's disagreement or differing interpretations), then make a decision you're comfortable with. You'll enjoy the experiences much more. It's all good!

[drseti]

I think it's important to differentiate between accidents and incidents. A few examples of incidents lumped together with accidents in the database:

A student pilot taxis too close to a hangar row, and breaks the wingtip strobe light. Incident.

A contractor parks a piece of machinery too close to a taxiway. Student pilot scrapes a wingtip on it, wingtip needs to be repainted. (This incident actually happened to my first SportStar.)

Another student pilot lands too close to edge of runway. Wheel hits and breaks runway light. Zero damage to aircraft, but because there was property damage, it's reported as an incident.

A pilot fails to fully latch the canopy, which comes open on takeoff. Pilot lands safely, closes canopy, and takes off again - but ATC files an incident report.

And finally, this one isn't in the database cited, but will be next year. It happened to me last month. I hit wake turbulence and get a cracked canopy. Declare an emergency and land safely. Incident.

Note that these are all non-injury, minor damage incidents. That they all involve Evektors is hardly a reflection on the quality of that particular aircraft, is it?

[SportPilot]

.......

[drseti]

Valid questions. Seems to be a rare occurrence. The only other ones I know about in the fleet were caused by bird strikes or canopies opening in flight. This one was a major turbulence event (fortunately rare). I have an EAA webinar on the incident coming up in May. Stay tuned.

[Cluemeister]

First of all Paul thank you for the response.

When you look at the weight argument, I and others took 4 pages of criticism for the suggestion that flying a certain plane at a certain weight may not be that unsafe. We were told it was back and white, and we were inviting both safety and legal consequences. Fair enough.

When it comes to real hard 10 year numbers that point to an alarming statistic about low wing safety, an exact reverse circle the wagons approach happens. Terms like the numbers are insignificant, each accident is different, and correlation does not equal causation are the rule of the day.

I believe a double standard exists because people don't want to believe that low wing planes are more likely to be in accidents. People may feel better explaining the data away, but that doesn't change the facts.

[SportPilot]

.......

[dstclair]

When it comes to real hard 10 year numbers that point to an alarming statistic about low wing safety, an exact reverse circle the wagons approach happens. Terms like the numbers are insignificant, each accident is different, and correlation does not equal causation are the rule of the day.

Clue -- your conclusion may be right or you may be wrong. The sample size is not large enough that the rates can be used as absolute comparisons. Google statistical significance and you'll find several tests that will help you validate the number. A standard p-value test will show that a finding a 1% could just have likely been 10% or .1%. This does not mean the rates are incorrect -- it just means they are not reliable. In other words, Flight Design's 'real' rate may indeed be 0.3% or it may be 10.3% since a null hypothesis of 10.3% cannot be dismissed. Likewise, the Sportcruiser may be 5.8% or it could be 0.8%.

Let's follow your logic further. The data was collected up until the end of 2014 and since that time there have been 3 FD fatal accidents and 1 SportCruiser. This would give the FD a rate of 0.8% and SC 1.1% over the last two years. An extrapolation of your logic would indicate that something could be horribly wrong with FD since they have increased there accident rate 3 fold. SC has gone down 80% so is now a better product. The truth is the data is not statistically significant so ANY conclusive can be supported.

The rest is left as an exercise for Paul's grad students

[drseti]

In the high vs. low wing debate, three separate questions have to be answered: what is the accident rate? What is the fatality rate? And what is the injury rate and severity? To draw statistacally significant * conclusions requires a large sample size, and there are just not enough LSAs out there to achieve significance. So, I suggest we widen our data set to include all GA aircraft. This question has been raised for at least the past 70 years, so there should be enough data out there to arrive at defensible conclusions.

* Statistical significance is a mathematical term, easily calculable, and not at all arbitrary. It computes, for a given sample size, the probability that a given result might have occurred purely at random. The Central Limit Theorem states that the larger the sample size, the more accurate the results. But when the entire population is small (as is the case in LSA accidents), even 100% sampling might not achieve statistical significance. That is why we sometimes have to expand the study to a larger population. For example, my early research into midair collisions did not achieve statistical significance because the entire population of MAC accidents is (thankfully) too small. So, instead I substituted near midairs (a much larger number than MAC accidents) as a proxy for MACs, to do a statistically significant analysis. See my 1990 book "Near Midair Collisions as an Indicator of General Aviation Collision Risk".

[drseti]

The data was collected up until the end of 2014 <snip> The truth is the data is not statistically significant so ANY conclusive can be supported.

<pedantic>
Dave, the data were collected up until... and are not statistically significant.
</pedantic>

[3Dreaming]

Something to remember is that the Czech Aircraft Works did not just make the Sport Cruiser. They also built the Zenith 601XL which had some in flight break up issues. Those accidents makes their number high, but not reflective of their current model airplane.

[Nomore767]

First of all Paul thank you for the response.

When you look at the weight argument, I and others took 4 pages of criticism for the suggestion that flying a certain plane at a certain weight may not be that unsafe. We were told it was back and white, and we were inviting both safety and legal consequences. Fair enough.

When it comes to real hard 10 year numbers that point to an alarming statistic about low wing safety, an exact reverse circle the wagons approach happens. Terms like the numbers are insignificant, each accident is different, and correlation does not equal causation are the rule of the day.

I believe a double standard exists because people don't want to believe that low wing planes are more likely to be in accidents. People may feel better explaining the data away, but that doesn't change the facts.

When you're evaluating the safety aspects of airplanes you also need to include the safety aspects of pilots.

The plane usually flies pretty well it just needs a qualified pilot to operate it. That pilot makes the decision on fuel requirements and if the weather is good enough for a safe flight. The airplane will let the plot fly it all the way to the scene of the accident.

GA and sport pilots tend to be less experienced and less skilled and I mean that in the kindest way. They haven't perhaps had military training, airline training, or acquired higher ratings which require more exacting skill levels and standards of judgement. They also fly less frequently and so may not be as proficient as they could be. If they're involved in accidents because of their own shortcomings that isn't the fault of the airplane, unless there is an obvious defect in the plane. This means that all pilots need good quality training, need to develop good judgement, experience and discipline, and need to fly regularly to hone their skills. Tailwheel pilots perhaps even more so.

There are some sport/GA pilots who just will not be told, and make casual decisions about fuel planning, weather and often tout that having a BRS chute installed will save them from themselves or if the situation deteriorates.
A recent FD CTLS accident off the shore of Long Island caused a fatality and whilst in fairness the accident report isn't out what is significant is that the pilot was flying pretty low, over water, at night in an airplane with a BRS chute installed and the airplane went into the sea. The BRS wasn't deployed so there was no benefit from its installation.
We don't know that there was any defect with the airplane yet but we do know that many other pilots would have perhaps advised against flying low at night over fairly large stretches of water in an SLSA. IMHO

Not casting aspersions as much as offering the point that the airplane may be 'safe' and the accident caused by other actors.

There is a pretty good old adage:-

"Aviation in itself is not inherently dangerous. But to an even greater degree than the sea, it is terribly unforgiving of any carelessness, incapacity or neglect."

[dstclair]

The data was collected up until the end of 2014 <snip> The truth is the data is not statistically significant so ANY conclusive can be supported.

<pedantic>
Dave, the data were collected up until... and are not statistically significant.
</pedantic>

whoops

[SportPilot]

.......

[BrianL99]

The data was collected up until the end of 2014 <snip> The truth is the data is not statistically significant so ANY conclusive can be supported.

<pedantic>
Dave, the data were collected up until... and are not statistically significant.
</pedantic>

Is the word "data" not generally accepted as a "mass noun", rather than a "count noun" ... which would allow the use of the singular, would it not?

[BrianL99]

Data are plural. Pie are round. Cornbread are square.

Data is not plural in English.

http://grammarist.com/usage/data/

http://www.onlinegrammar.com.au/top-10- ... ural-verb/

http://www.theguardian.com/news/datablo ... l-singular