Copious, possibly annoying, warning messages notwithstanding, it appears that,
no matter which way a software package goes on the issue, you need to stay
cognizant of how your package is handling missing values or you will get
burned.
Joseph Coveney
--------------------------------------------------------------------------------
From: Quentin McMullen ([email protected])
Subject: Re: Suggestions made at SUGI
Newsgroups: comp.soft-sys.sas
Date: 2004-06-04 06:15:03 PST
I agree completely with your suggestion, Jack.
I've seen folks get horribly burned by tabulate. If a table has multiple
class variables, e.g (table race haircolor eyecolor, gender), tabulate
happily excludes the entire record from the table if just one class
variable is missing. Seems sinful to me. A system option would make me
much happier than having to remember to add "missing" to every proc
tabulate.... Or like you say, at least have a NOTE: I could catch.
Basically in the tabulate setting, it's equivalent to the NOTE that
says "The graph printed excludes data that was outside of the defined
axes."
Kind Regards,
--Quentin
On Thu, 3 Jun 2004 11:39:42 -0500, Jack Hamilton
<[email protected]> wrote:
>The question is, then, why is it the default behavior of many base SAS
>procedures to silently throw away observations with missing class
>variables? I could understand (but not like) throwing them away and
>writing a message to the log, but why throw them away silently?
>
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
