Chronology Current Month Current Thread Current Date
[Year List] [Month List (current year)] [Date Index] [Thread Index] [Thread Prev] [Thread Next] [Date Prev] [Date Next]

Re: precision?



At 07:47 AM 9/1/99 -0400, Ed Eckel wrote:
Does it strike anyone else as, well, disturbing that discussion on this
list reveals so many variations on what is meant by 'precision' or
'accuracy'?

It doesn't disturb me.

Physics may be a science, but teaching remains an art. Teaching,
especially at the introductory levels, requires tremendous artistic
judgement about what simplifications to make.

Characterizing the uncertainty in a measurement is a very complex subject.
Introducing the subject by stating simplified definitions for three or four
terms -- terms that were qualitative and informal to begin with -- has some
advantages but involves considerable risks and tradeoffs.

IMHO, these qualitative, informal terms have a rather limited range of
applicability. For low-quality measurements, the distinctions between
accuracy / precision / resolution don't matter very much. For high-quality
measurements, you need to worry about a lot more than that.

The pros typically just talk about "uncertainty" and give a detailed
analysis of the many contributions to the uncertainty.

As it says on
http://physics.nist.gov/cuu/Uncertainty/international2.html

% The end result of the work of ISO/TAG 4/WG 3 is the 100-page _Guide
% to the Expression of Uncertainty in Measurement_ (or GUM as it is
% now often called)

As it says in
http://www.cornnet.nl/~mlbroens/vim.htm
% "Accuracy" is a qualitative concept.

I cannot resist paraphrasing this as
"accuracy" is not a precise term.

and in turn that reminds me of the old saying
There is no news in _Pravda_, and there is no truth in _Isvestia_.
but we digress.....

=====================

If we have not all been trained to some consensus on the
meaning of various terms that we teach, perhaps it is not surprising
that researchers are able to easily "show" that our students don't know
physics.

Are you worried that they don't know certain bits of terminology, or don't
know the concepts? Lois's .38 caliber demo should give the students a
well-nigh unforgettable lesson in the concepts of width of distribution,
center of distribution, the iterative nature of experiment design, et cetera.

And just because "accuracy" is not a precise term, doesn't me we shouldn't
use it or that we shouldn't teach it. We routinely use English and other
natural languages to describe qualitative and approximate concepts. We
have another language that we use when we want to be really, really precise
-- the language of mathematics. For the .38 caliber example, we can take
refuge in
-- mean and standard deviation, or better yet
-- first moments and second moments
-- correlation coefficients
-- et cetera.

My personal taste would be to start with a relatively general discussion of
probabilities, and then treat data analysis as a special case thereof.
Your mileage may vary.