[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]
st: RE: retry: copy value withing group
-bysort id (x): replace x = x-
(Or you ou could use Tony Brady's -xfill- , - net from
http://www.sealedenvelope.com/ - to locate and download)
> -----Original Message-----
> From: email@example.com [mailto:owner-
> firstname.lastname@example.org] On Behalf Of email@example.com
> Sent: Monday, October 23, 2006 8:46 AM
> To: firstname.lastname@example.org
> Subject: st: retry: copy value withing group
> My appologies. The first several lines of my previous e-mail were
> mysteriously deleted. Hopefully this one is complete.
> *Intro*: Here's a simple Stata task I solved years ago and do
> frequently, yet I have never been happy about how I do it. I've always
> wondered if there was another way.
> *Fact*: In my data there exists a variable, x, with the following
> property (known by construction or verified with an -assert-). Within
> each group of observations identified by a unique id (call the variable
> id), x is missing for all but one observation.
> *Operation*: I want to assign the single nonmissing value of x to all
> missing values of x within the group. That is, when I'm done, x is
> nonmissing for all observations and constant within id. This constant
> value of x (within id) remains unchanged for the original nonmissing
> values of x.
> *My Solution*: Usually I accomplish this with something like:
> bysort id: egen t = mean(x) //other -egen- commands work, like
> -max()-// replace x = t
> drop t
> *Comment*: This works because I choose an -egen- command that ignores
> missing values. What displeases me about this is that when I review my
> code later, it is hard to tell what I'm really doing since my
> motivation is not to compute a mean or max, per se, but just to execute
> a sort of copy operation within group. What would be nice is if there
> were a command named copy() (or something clear) that (1) verified the
> properties I require (see "Fact" above) and then (2) did the operation
> I want (see "Operation" above). Even a more transparent way of doing
> (2) (and not (1)) would be an improvement. I'm not skilled enough to
> easily write such a command. Perhaps there is such a command or some
> other way to do what I want that is more transparent than my solution.
> Any suggestions?
> * 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: