How safe is flying REALLY?

The logic you are using to interpret my assertation that (basically)
statistics (including air safety) are not absolute, and subject to
interpretation & manipulation is confusing.

I'm not saying safety doesn't exist at all. What I AM saying, is you can
make just about anything look just how you want by using the right method.



is depends on who you ask & how they tell you.

That's tantamount to saying that nothing is any degree of "safe", from
which
we would conclude that nothing is safer or less safe than anything else.

And that's clearly false, so I must reject your statement.

In fact everything has some degree of safety, and we can argue about how
to
most accurately and usefully measure it, but I don't think we can throw
out
the question.

miguel
--
Hit The Road! Photos and tales from around the world: http://travel.u.nu

The logic you are using to interpret my assertation that (basically)
statistics (including air safety) are not absolute, and subject to
interpretation & manipulation is confusing.

I'm not saying safety doesn't exist at all. What I AM saying, is you can
make just about anything look just how you want by using the right method.



is depends on who you ask & how they tell you.

That's tantamount to saying that nothing is any degree of "safe", from
which
we would conclude that nothing is safer or less safe than anything else.

And that's clearly false, so I must reject your statement.

In fact everything has some degree of safety, and we can argue about how
to
most accurately and usefully measure it, but I don't think we can throw
out
the question.

miguel
--
Hit The Road! Photos and tales from around the world: http://travel.u.nu

Flying remains, statistically speaking, the safest means of transportation.

Unbeknown to many, the majority of large airlines use Israeli security
services.

Pay attention to your fellow travelers, what they carry on board, and remember
that it's better to call in a tip and be wrong, than to be embarrassed to do so
and regret it later.

Regards

Travel Maven
http://www.travelandlearninc.com

Flying remains, statistically speaking, the safest means of transportation.

Unbeknown to many, the majority of large airlines use Israeli security
services.

Pay attention to your fellow travelers, what they carry on board, and remember
that it's better to call in a tip and be wrong, than to be embarrassed to do so
and regret it later.

Regards

Travel Maven
http://www.travelandlearninc.com

(Reef Fish) wrote in message . com...
[snip]
I was trying to make several points all at once, and did not make it
sufficiently clear that apart from the fact that it takes YEARS of
training to analyze data well, the students at the HS level can
actually learn all there is to learn in ELEMENTARY CALCULUS and be
able to apply the calculus concepts to daily, applied problems, while
that is quite impossible in the subject of statistics because there
is so much subtle in the subject and so little 100% right or wrong
answers (as in calculus).
[snip]

I think the largest problems in the general publics understanding
of statistics have little to do with the nuances of statistics and
everything to do with how data is collected and analyzed. The number
one complaint I have about an awful lot of data analysis that is
presented is any kind of "control group". I am constantly being
presented with numbers that "sound bad" but with which I have
no ability to judge magnitude. A favorite of mine was some data
that came out about how "dangerous" communter airline flights were.
The data was presented in comparison to longer flights by "majors".
The problem was that it was all normalized to flight time. Take
off and landing are two of the more dangerous times. And both
kinds of flights spend almost the exact same amount of time
in a take off and landing mode. Furthermore, they didn't compare
flying to DRIVING between the two cities and if that was done,
often driving was more dangerous than flying. Folks were eschewing
flying on commuters to drive instead. Blitheringly stupid.

The second largest issue, and one common to the airline safety issue,
is the sensitivity to a single data point. It is related to "statistically
significant sample" but basically it is a condition where any single
data point can make huge variations in the statistical outcome.
Commercial, scheduled airplanes crash, with fatalities, so infrequently that
any single crash can make large variations in a single years statistics.
A simple piece of information that any news article could provide, along
with standard deviation which is NEVER reported except in polling data, is
the magnitude of effect of the most influential data point.

(Reef Fish) wrote in message . com...
[snip]
I was trying to make several points all at once, and did not make it
sufficiently clear that apart from the fact that it takes YEARS of
training to analyze data well, the students at the HS level can
actually learn all there is to learn in ELEMENTARY CALCULUS and be
able to apply the calculus concepts to daily, applied problems, while
that is quite impossible in the subject of statistics because there
is so much subtle in the subject and so little 100% right or wrong
answers (as in calculus).
[snip]

I think the largest problems in the general publics understanding
of statistics have little to do with the nuances of statistics and
everything to do with how data is collected and analyzed. The number
one complaint I have about an awful lot of data analysis that is
presented is any kind of "control group". I am constantly being
presented with numbers that "sound bad" but with which I have
no ability to judge magnitude. A favorite of mine was some data
that came out about how "dangerous" communter airline flights were.
The data was presented in comparison to longer flights by "majors".
The problem was that it was all normalized to flight time. Take
off and landing are two of the more dangerous times. And both
kinds of flights spend almost the exact same amount of time
in a take off and landing mode. Furthermore, they didn't compare
flying to DRIVING between the two cities and if that was done,
often driving was more dangerous than flying. Folks were eschewing
flying on commuters to drive instead. Blitheringly stupid.

The second largest issue, and one common to the airline safety issue,
is the sensitivity to a single data point. It is related to "statistically
significant sample" but basically it is a condition where any single
data point can make huge variations in the statistical outcome.
Commercial, scheduled airplanes crash, with fatalities, so infrequently that
any single crash can make large variations in a single years statistics.
A simple piece of information that any news article could provide, along
with standard deviation which is NEVER reported except in polling data, is
the magnitude of effect of the most influential data point.

(me) wrote in message . com...
(Reef Fish) wrote in message . com...
[snip]
I was trying to make several points all at once, and did not make it
sufficiently clear that apart from the fact that it takes YEARS of
training to analyze data well, the students at the HS level can
actually learn all there is to learn in ELEMENTARY CALCULUS and be
able to apply the calculus concepts to daily, applied problems, while
that is quite impossible in the subject of statistics because there
is so much subtle in the subject and so little 100% right or wrong
answers (as in calculus).
[snip]

I think the largest problems in the general publics understanding
of statistics have little to do with the nuances of statistics and
everything to do with how data is collected and analyzed.

Correct. One very relevant factors:

The number one complaint I have about an awful lot of data analysis
that is presented is any kind of "control group".

This can be further broken down to cases (1) "control groups" CAN
be carefully selected in a "designed" experiment; (2) "control
groups" CANNOT be designed in certain cause-effect studies.

A hypothetical example: If you want to study the beneficial/harmful/
or whatever effect of smoking dried banana peels (as opposed to tobacco,
say), there does not exist any true CONTROL/EXPERIMENTAL groups (for
ethical reasons) that you take a randomly chosen group of otherwise
comparable individuals and force them to smoke banana peels, or
tobacco (against their will) to determine how harmful the effects
are!

So, folks use RATS, and then force rats to smoke the equivalent of
100,000 cigarrettes a day (compared to human), and then pretend the
study on rats apply to humans. :-)



The second largest issue, and one common to the airline safety issue,
is the sensitivity to a single data point.

From a statistical point of view, that is primarily the DEFECT of
the branch of statistics that is generally known as the "classical"
or "frequentist" approach. In another ever-increasingly used
branch of statistics, especially in the biomedical area, is that of
Bayesian statistics, where every single data point carries its own
weight, and statistical conclusions CAN be drawn on ONE data point,
integrated with the expert opinon (quantified via something called
a "prior distribution"). The latter also explains, logically and
scientifically, why a single piece of data, or a single collection
of the same data, have DIFFERENT impact on various people with
different expertise on the problem.


But these are just the tip of the iceberg on the field of statistics,
whose intricacies are known or understood by very few; but whose
abuses are happily applied or accepted by the majority!


That's the way it is; and will be for years to come. Trust me.


-- Bob.