[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: RE: missing values included in conditions
This one is not weird at all (in Stata terms).
Missing is not zero. That is the beginning and end of it.
Considered as inputs, in Stata,
not 0 <-> true
0 <-> false
Considered as outputs, in Stata,
1 <-> true
0 <-> false
That is all you need to know, but the story is explained at excruciating length in
FAQ . . . . . . . . . . . . . . . . . . . . . . . True and false in Stata
2/03 What is true and false in Stata?
The alternative would be some three-valued logic, in which missing was something
else (or ignored). Many people would prefer that, but even those who would prefer that
can't seem to agree what it should be. Clearly there are many contexts in which
missings are ignored, but this isn't one of them.
At the first Boston users' meeting, David Kantor gave a talk on one
possibility, which provoked a very lively (but inconclusive) discussion.
Fortuitously, that was followed by a talk on circular arguments.
> And another weird thing I just noticed and wanted to confirm
> that it's my
> mistake not Stata's :)
> lets suppose we have a variable cond
> if I write
> .. count if cond==1
> it will of course return 2. But using
> ..count if cond
> gives result 3, as Stata includes missing in count as well.
> Somehow I was
> thinking that these two approaches were equivalent (and
> missings are always
> excluded) but was proven wrong.
> So should missings be treated as "True" in this kind of conditions?
* For searches and help try: