Rate of fatal accidents in Mooneys over time

[EricJ]

1 hour ago, Ragsf15e said:

I agree statistics and testing are a powerful tool, but I’m just finishing my masters in economics, and the last class I had on econometrics was disheartening.  There are so many pitfalls, methods, tests, choices, and grey areas that it seems like two things happen… either the researcher can make the end result say what they like (to a certain extent) or someone with less experience will just muff a very complicated situation and draw the wrong conclusions.  Omitted variable bias and selection bias are easy ones to understand and see how they screw up everything.  In the end, I feel like it’s the experience and impartiality of the researchers as much as the data they have.

This is pretty much true in all branches of science, math, and engineering, in my experience, and my career was essentially as an applied mathematician for communications engineering using lots and lots of stochastic (statistical) models.    Statistics are highly valuable and useful and revealing when used correctly, but, like anything else, can be corrupted or misused or misunderstood or misinterpreted.

In my career if the output of a statistical analysis was needed to make a useful decision on how to proceed in a research project or engineering decision, there was always a lot of verification of models and cross-checking or some means of validation that it made sense.    This was using underlying models that had similar verification or validation credentials so that it was known that they could be relied on in the particular application, i.e., that any limitations or shortcomings of the models had been accounted for.    From my limited exposure this is also true in economics, but the assumptions and limitations of models are sometimes stretched because there may be political or economic motivation to do so, or hubris or whatever.   There's also often a harder time collecting relevant or useful data, which is critical for a useful output.

Twain was a humorist, so while there is some wisdom in his comments they should be taken fwiw.

[AndreiC]

I am surprised no one has tried to figure out hours flown by Mooneys (or any other make/model) from ADS-B data like flightaware. I don’t know if they let you read their data, but it should not be too hard to write a script to get what you want if they let you do it. 

[0TreeLemur]

4 hours ago, AndreiC said:

I am surprised no one has tried to figure out hours flown by Mooneys (or any other make/model) from ADS-B data like flightaware. I don’t know if they let you read their data, but it should not be too hard to write a script to get what you want if they let you do it. 

Some percent of M20s don't have ADS-B out capability.   Untangling that will take some effort too, as it adds uncertainty to fleet annual flight time.

[AndreiC]

2 hours ago, 0TreeLemur said:

Some percent of M20s don't have ADS-B out capability.   Untangling that will take some effort too, as it adds uncertainty to fleet annual flight time.

But every single fatal accident I have read about had tracks on flightaware. So maybe this is not so unreasonable for recent years, planes which don't have ADS-B probably sit.

[0TreeLemur]

25 minutes ago, AndreiC said:

But every single fatal accident I have read about had tracks on flightaware. So maybe this is not so unreasonable for recent years, planes which don't have ADS-B probably sit.

@Hank?

[Hank]

3 hours ago, 0TreeLemur said:

Some percent of M20s don't have ADS-B out capability.   Untangling that will take some effort too, as it adds uncertainty to fleet annual flight time.

53 minutes ago, AndreiC said:

But every single fatal accident I have read about had tracks on flightaware. So maybe this is not so unreasonable for recent years, planes which don't have ADS-B probably sit.

27 minutes ago, 0TreeLemur said:

@Hank?

My C has zero ADS-B capability. When I travel, especially IFR, i generally show up on FlightAware. When I'm out putting around squawking 1200, of course I don't. Traveling one or two states away VFR with Flight Following, there is onkymsometimes a record and I've not figured out the pattern yet.