[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
RE: st: RE: resticting the sample?
I haven't got any gun in my hand, but
I'm not clear that we need this syntax.
Are you asking for what can be done by
... if !inlist(_n,10,100,250)
If that gets tedious, a macro assignment
cuts the typing
local outliers "inlist(_n,10,100,250)"
regress <whatever> if !`outliers'
Note that this could bite you. Sort
the data and the observation numbers
refer to different observations, but
then -suspendin()- as I understand it suffers from
exactly the same problem. So I go
back to my earlier suggestion
to -generate- an in-sample indicate
variable, or the other suggestions made.
See also -inrange()-.
> Dimitriy V. Masterov replied to Shuaizhang Feng:
> > Your can use preserve/restore to do this:
> > preserve
> > keep if age<40
> > reg ....
> > restore
> And Nick Cox replied to Shuaizhang Feng with:
> > gen mysample = (age < 40) & (nobel_prize == 1) &
> > == 1)
> > ...
> > reg ... if mysample
> > See also http://www.stata.com/support/faqs/data/trueorfalse.html
> Running regressions using restricted samples (or by leaving out
> disqualifying observations) is something I've been really
> keen to do for
> quite some time. But solutions such as the ones above seem to
> be the only
> ones available (but that doesn't mean to say that they're bad).
> I'm not really one for adding to StataCorp's burden, for they probably
> have enough to do in preparing for version 9. But wouldn't it
> be great if
> one could run a regression in the following way:
> . reg tomatoes potatoes onions sausages, suspendin(10 100 250)
> so that you could simply suspend the observations without
> having to -drop-
> them for that regression and then have to -restore- the dataset
> afterwards? Desirably, this option would automatically pick up the
> observation numbers as they appear by row in the data editor.
> No doubt this idea will be shot down in myriad ways and
> people would say
> that it couldn't work. But, to quote the British comedy duo
> Armstrong and
> Miller, "Yes, but it would be good if it did."
> CLIVE NICHOLAS |t: 0(044)191 222 5969
> Politics |e: firstname.lastname@example.org
> Newcastle University |http://www.ncl.ac.uk/geps
> * 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/
* For searches and help try: