This solution of directly counting below, at and
above is exactly what sprang to my mind too.
As Nick would acknowledge, his code
is not industrial strength, one reason being
the need to count missings too.
Nick
[email protected]
Nick Winter
> Assuming your variable is mpg, and your scalar is 20. Then how about
> something like:
>
> count if mpg>20
> local a = r(N)
> count if mpg<20
> local b = r(N)
> count if mpg==20
> local e = r(N)
>
> di (`b'+`e'/2) / (`a'+`b'+`e')
>
> That is, count how many observations are below s, how many
> above, and how
> many are equal. Then the percentile is just the number
> below, and half of
> the ties, over the total N.
> >Hi, I have a question regarding percentiles that I've been
> having some
> >trouble with...suppose that you have a variable x, and you
> have a scalar
> >s, and you'd like to know at about what percentile of the x
> variable the s
> >scalar falls. For instance, suppose x=1,2,3,...,1000. Then
> if you have
> >s=20, it should be assigned the percentilie value of the 2nd
> >percentile. Is there a way to do this easily in stata? (I wanted to
> >assign the scalar a n approximate percentile EVEN IF IT'S
> NOT A VALUE in
> >the variable itself so it can't be located in there)
> Everything I could
> >think of would take quite a while.
*
* 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/
