I pick the "amiable" happily; I don't detect
any insults in this thread, just a miscellany of comments,
several of which seem irrelevant to what I had to say.
"Consistency" is the answer, in one word. If
one number sorts higher than another, it counts
as greater than it. And if people want to say
that is relevant in one context and not
another, they could well be right. But all
code has to be written to
do what I say
whereas users still seem to expect that
a program understands this somehow as
do what I mean
I am happy to learn that SPSS implements
this approach, so that it is readily available
to those who want it. I am reminded of Dennis Ritchie's
canonical comment when asked why C was as spare
as it was, "If you want PL/I, you know where to
find it".
StataCorp no doubt welcome all code fixes
that make it impossible to do silly things.
Nick
[email protected]
> -----Original Message-----
> From: [email protected]
> [mailto:[email protected]]On Behalf Of
> R. Allan Reese
> Sent: 08 January 2004 14:18
> To: [email protected]
> Subject: RE: Missing as true - Was: Re: st: RE: Another
> Stata feature
>
>
> Since we are trading amiable insults here, I'll suggest it
> is Nick who
> has confused the principle of processing data with the computational
> problem of representing the data. I certainly have no qualms about
> teaching users that something may be TRUE, FALSE or
> UNKNOWN, and Nick's
> reluctance reinforces my view that politicians and managers
> are barking
> mad when they always demand (or claim to know) "the answer" with
> "absolute certainty".
>
> Stata implements value+missing -> missing
> value*missing -> missing
> and
> value / 0 -> missing
> so why should
> value>0 -> 1, when value is missing
> be a sacred cowpat?
>
> I agree there are problems in now changing the definition, so have
> suggested a warning in the output. Otherwise you *will* have
> computer-aided disasters due to mixing missing and real
> data. I wonder
> how many medical analyses using Stata have used something like:
> regress y x1-xN if age>60
>
> I'll shut up on this one.
> Allan
>
> Oh the string variable was only specified because "invade 5" didn't
> seem to make any sense whereas invade "Iraq" ... damn, maybe that
> doesn't make any sense either.
*
* 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/
