SCUBA-SE Archives

September 2000

SCUBA-SE@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Reef Fish <[log in to unmask]>
Reply To:
SouthEast US Scuba Diving Travel list <[log in to unmask]>
Date:
Thu, 21 Sep 2000 22:16:42 -0400
Content-Type:
text/plain
Parts/Attachments:
text/plain (50 lines)
On Thu, 21 Sep 2000 16:48:12 -0800, Kent Lind <[log in to unmask]> wrote:

>Bob, maybe you can tell us about the difference between accuracy and
>precision and which of the two concepts is  most relevant when determining
>how many decimal points to use on a measuring instrument.

"Accuracy" and "precision" are closely related concepts.  They are both
relevant when determining how many decimal places to use.

This is best illustrated by the particular examples in question, UBS
and Handi.  Strike (it's HIM this time <G>) said about the UBS,

DS>  They're apparently accurate to one tenth of one per cent

That .1% is both the accuracy and precision for which the UBS is capable.

To present a reading as 32.12345 by the UBS would be a falsely-precise
but inaccurate reading.  That number should IMPLY a PRECISION to the
5th decimal place.  Otherwise, the digits 2345 are meaningless because
it is accurate to ONLY 1 decimal place.

On the other hand, if UBS shows a calibration of 32.1 as 32, or round
it to only ONE nonzero digit, 30, then the reading would be both
inaccurate and imprecise, given the precision of the instrucment.

The number of "significant digits" used in statistics and numerical
analysis of DATA should (and often does) IMPLY the precision of the
measurement.  Thus, when the number of psople in the armed forces at
some country at some time is given as 538,000 it is clearly understood
that there aren't EXACTLY that many bodies, but some number between
537,500 and 538,499.  Conversely, a figure of 528,321 given for any
particular year is clearly false-precision.


It is in THIS CONTEXT that the Handi's calibration is no more accurate
than the unit digit, such as 32;  and it makes no sense to claim (or
imply) that it is MORE accurate than the UBS which is accurate to 1/10
of one unit of measurement (I am avoiding the use of "percent" because
it has two different meanings in this precision context.


Which decimal should you use?  The ACCURACY presceibed by the spec of the
instrument of course.  Not that you have to USE that precise value
necessarily, but why use a fuzzy number when you have a non-fuzzy one.

You may send $100.00 to your favorite charity as your tuition for this
requested mini-lecture.  :-)

-- Bob.

ATOM RSS1 RSS2