Date: Wed, 8 Dec 1999 00:37:41 +0000
Reply-To: John Whittington <John.W@MEDISCIENCE.CO.UK>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: John Whittington <John.W@MEDISCIENCE.CO.UK>
Subject: Re: When large number of significant digits may be required
In-Reply-To: <199912070216.CAA14667@vicar.netnames.net>
Content-Type: text/plain; charset="us-ascii"
At 00:35 06/12/99 -0500, Hermans1 wrote:
> Was it Oskar Morgenstern in On the Accuracy of Economic
> Observations who estimated that results of calculations of the US
> national debt have a level of precision at no better than the
> $100,000. level? I contend that statisticians and programmers
> have an obligation to resist pointless, misleading, and wasteful
> pretense of precision. Sig
Indeed - but they also need to be sensible about how they 'manage' whatever
degree of precision they deem appropriate. I am, in particular, reminded
that rounding to an appropriate (and realistic) number of significant
figures generally should only be done *once*, at the end of all the
calculations - which themselves should utilise data to the full extent of
whatever precision is available.
I have seen some very serious errors resulting from 'repeated stages of
rounding' or, perhaps most disasterously, when a high degree of rounding is
applied to 'source' figures, millions of which are to be combined to
produce the final answer - you may only want, say, the 'total income of all
US citizens' expressed to 'the nearest $100,000' - but the person who
decides to apply that 'rounding' at the start, and reduce all individual
incomes to 'the nearest $100,000' is in for a bit of a shock :-)
... just my two pence worth!
Regards,
John
----------------------------------------------------------------
Dr John Whittington, Voice: +44 (0) 1296 730225
Mediscience Services Fax: +44 (0) 1296 738893
Twyford Manor, Twyford, E-mail: John.W@mediscience.co.uk
Buckingham MK18 4EL, UK mediscience@compuserve.com
----------------------------------------------------------------
| 
