Date: Tue, 5 Feb 2002 22:54:57 +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: Missing values for character variables
In-Reply-To: <E16YE6Z-0004gt-00@relay1.netnames.net>
Content-Type: text/plain; charset="us-ascii"; format=flowed
At 17:30 05/02/02 -0500, Quentin McMullen wrote:
>As far as I know, there is nothing for character variables analogous to
>the special missing values for numeric variables. I can certainly see how
>this would be useful.
>I guess one problem is how to represent a "character special missing" (in
>code and internally) with some out-of-domain value. For numeric variables
>.A-.Z are clearly out of the domain of numeric values.
>
>Any thoughts on what could constitute a good character special missing
>value system? How hard would this be for SI to institute?
Yes, it would often be useful (probably at least as often as with numeric
variables) - but I think you've defined the difficulty very well. Such a
facility would presumably be possible if some (necessarily 'printable')
character(s) were defined as being 'out-of-domain' for SAS string variables
- and I suspect that there would 'always be someone' who had a problem in
relation to whatever character they had been denied! As things are, as
others have said, the use of a format seems to be the most obvious
workaround. I suppose an alternative (but rather wasteful) method would be
to associate a second (probably numeric) variable with each of the string
variables of interest, and use that secondary variable to carry information
on 'types of missing data'.
Kind 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
----------------------------------------------------------------
| 
