--- On Tue, 1/5/10, Peter Flom <peterflomconsulting@mindspring.com> wrote:

> From: Peter Flom <peterflomconsulting@mindspring.com> > Subject: Re: New Comparison Operators? - WAS: missing numerical values = - > To: "Dale McLerran" <stringplayer_2@YAHOO.COM>, SAS-L@LISTSERV.UGA.EDU > Date: Tuesday, January 5, 2010, 10:20 AM > Dale McLerran <stringplayer_2@YAHOO.COM> > wrote > > > >If you don't like Lou's example, how about missings due to > >a skip pattern? > > > >if X = "skipped smoking question because person is nonsmoker" and > > Y = "skipped smoking question because person is nonsmoker" > > then x does equal y. > > > > > >That is certainly a valid comparison. Here, if we code X=.S > >and Y=.S, then I would expect the comparison "if X=Y" to > >return as TRUE. > > > >This is not to argue that the way that SAS has implemented > >missing values is the "right" approach. However, it does > >argue that there is validity of the SAS approach for at least > >some problems. Further, the approach of programs like R is > >not always appropriate. > > > > Hi Dale > > Yes, you're right > > In R, here, you'd have to do something else. > > So, neither default is always ideal, which indicates that > it would be nice to have both possibilities, doesn't it? > > Peter > > Peter L. Flom, PhD > Statistical Consultant > Website: http://www > DOT statisticalanalysisconsulting DOT com/ > Writing; http://www.associatedcontent.com/user/582880/peter_flom.html > Twitter: @peterflom >

Well, yes and no. With options comes complexity. How do you implement both? How do you determine which rules will be enforced for a particular comparison. SAS documentation of handling of missing values must of necessity become more complex. Will people who are not experienced pay attention to all of the issues and complexity of the different methods of handling missing values?

Remember, too, that it is issues like this which result in software becoming bloatware. When the software is modified to provide some special handling that was not originally intended, it can be difficult to incorporate the new feature(s) in well structured code. This can result in longer code which runs slower.

Dale

--------------------------------------- Dale McLerran Fred Hutchinson Cancer Research Center mailto: dmclerra@NO_SPAMfhcrc.org Ph: (206) 667-2926 Fax: (206) 667-5977 ---------------------------------------