Scheme is according to taste a variant, relative or
dialect of Lisp with "lots of irritating superfluous
parentheses", as one story goes. Stata is in the lineage
of C and so on the whole its constructs are not similar
to those of Scheme. (Similar story for S.)
In this particular case you want observations that
have values that are not missing on all the variables
in a list. And you also asked about -egen-. As you
are aware of -egen-, you could also have looked at
the help to discover pertinent functions. The easiest
is probably -egen, rowmiss()-.
egen nmiss = rowmiss(med*)
counts missings across a varlist so that the condition
you want then becomes
if !nmiss
as -nmiss- of 0 means present on all variables, -nmiss-
of 1 or more means missing on at least 1 variable and
negation flips true to false and vice versa.
If -egen- did not exist, you would not really need
to reinvent it.
gen anymiss = 0
quietly foreach v of var med* {
replace anymiss = anymiss | missing(`v')
}
gets you where you want to be (logically) so
that
if !anymiss
is then an appropriate condition. The same kind
of construct could be used with the operator &
for other problems. Similarly the
guts of -egen, rowmiss()- are, applied to your
variables,
gen nmiss = 0
quietly foreach v of var med* {
replace nmiss = nmiss + missing(`v')
}
David Kantor's module -trinary- is intriguing
stuff, but it seems to me that your problems
require only official Stata.
Nick
[email protected]
David Kantor
> I don't think you can "map" an arbitrary operator.
>
> But there are many ways to program your problem.
>
> If you create a set of indicators as to whether the med variables are
> nonmissing (and by the way, a more generally correct test is
> ~mi() )...
> gen byte nonmis_med1 = ~mi(med1)
> gen byte nonmis_med2 = ~mi(med2)
> etc., (and that can be made into a loop)
>
> then if you have the trinary module, you can do
>
> egen byte med_ok = rtvor(nonmis_*)
>
> this will give you the row-or of the arguments. (The fact that it is
> 3-valued logic is irrelevant.)
> Then you can
> list med* if med_ok
>
>
> To get trinary,
> ssc inst trinary
Joel J. Adamson
> >Does anyone know how to map a logical operation (e.g., | or &) across
> >a list so that I can save myself some typing?
> >
> >Let me give you an example (the one that prompted the search for an
> >answer):
> >
> >I want to list a set of variables, only if an observation
> is not missing
> >for the variables in question. We usually handle this by
> constructing
> >a small logical expression:
> >
> >l med1 med2 if (med1~=.|med2~=.)
> >
> >However, if I have six, or a hundred "med" variables (med*), it would
> >be easier (less error-prone) to type:
> >
> >l med* if |(med*~=.)
> >
> >Where the "|" before the "(" maps the operation onto the
> expanded list
> >of variables.
> >
> >I know that Scheme (and hence probably S) have a function
> called "map"
> >that maps any operation onto a list, e.g.,
> >
> >(map or '(1 2 3 ... ))
> >
> >Does anyone know of a similar construct in Stata? How about an egen
> >function that would accomplish the same goal?
*
* 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/
