RE: More About Those Trucks...(Alfa content=0)

[Chris.Prael@domain.elided]

From: Carlo.Cardilli@domain.elided

> After the safety improvements the rate of serious injury 
> per miles driven remains the same (statistically speaking, 
> so skeptics can always argue about this).

I have a BSc. in math, which included a few classes in applied measure
theory (statistics) and some experience in using statistics in
engineering applications.

One of the first issues with "statistics" is to understand the
difference between a statistic and an estimate (wild ass guess).  A 
statisitc is compiled from a collection of measurements.  Only 
measurements can go into a statistic.  So, if any value that has not 
been measured in incorporated into the final value, the result is not 
a statistic, it is a WAG.

If you think about the paragraph above a little, it becomes clear that
very few of the numbers that are published by various authorities in
the United States on automotive safety are statistics.

Carlo's paragraph, above, illustrates this problem very well.  We can
assume that a credible count of "serious" injury can be made, and has
been.  (Though we should require a rigorous definition of "serious"
before we accept such data.)  On the other hand, although it is not 
impossible to measure the number of miles driven in a period of time, I 
can guarantee that noone has ever done so because the cost is 
prohibitive.  

So, someone has made a WAG, possibly based on some measured value or 
other,  and called it a 'statistic'.  It isn't! We do know that the 
amount of fuel sold is measured fairly accurately.  While this is NOT a
measure of fuel consumed, one can guess that the amount of spillage,
evaporation, and other sources of non-use is consistent.  The safety
authorities then estimate a national average fuel consumption rate and
use that to "compute" the number of vehicle miles.  Obviously, anything
computed from a number derived as this one is cannot be legitimately
called a 'statistic'.

It is worth noting that, unlike the US government, European governments 
publish real statistics.  They divide a measurement (say collisions) by
the number of registered drivers.  When this measure is used, US
drivers are a bit less safe than Spanish drivers and a lot worse than
German and British drivers.

Chhris Prael

------------------------------