Re: certainty versus uncertainty

[Bob Sciamanda <Sciamanda@worldnet.att.net>]

LUDWIK KOWALSKI wrote:
> My wife asked me, "how sure are you that this was done correctly"? The
> answer was "about 95%". By implication I was saying that my uncertainty
> was only 5%. There is an interesting assymetry here. The difference
> between being 95 versus 99 percent sure is about 4% while the difference
> between being 5 versus 1 percent uncertain is 500%.
>
> Nothing profound, just an observation. Which way of answering is better,
> in terms of certainty or in terms of uncertainty?
>                                                      Ludwik Kowalski
Food for flames:
Relative deviations or errors (ie., [delta x]/x ) are meaningful in
reporting physical measurements because the absolute value being
measured is relevant - the relative error makes explicit that an error
of .1gm is much less significant in the measurement of a 1 kg object
than it is in the measurement of a 1 gm object.

This does not apply to the type of "measurement" which your wife is
asking you to make, ie., the relative error in your "measurement"
(however you report it) would artificially inflate deviations from low
absolute values, even though your "measurements" are equally just
guesses no matter where you are on the scale. 
-- 
Bob Sciamanda		sciamanda@edinboro.edu
Dept of Physics		sciamanda@worldnet.att.net
Edinboro Univ of PA	http://www.edinboro.edu/~sciamanda/home.html
Edinboro, PA		(814)838-7185