I am not sure whether Olexandr's solution is what Max R is seeking, but
I want to clear two marginal issues out of the way.
-gsort- is not a user-written command and any attempt to install it from
SSC will not work. It is part of official Stata, and has been since it
was written.
The test
... if indep3 == ""
presupposes that -indep3- is a string variable, which seems unlikely.
But more general code that works whether -indep3- is string or numeric
is
... if missing(indep3)
Incidentally note that
sort panelid timevar
by panelid: replace indep3 = indep3[_n-1] if missing(indep3)
can be slimmed to one command
bysort panelid (timevar): replace indep3 = indep3[_n-1] if
missing(indep3)
There's a more detailed and systematic discussion in
FAQ . . . . . . . . . . . . . . . . . . . . . . . Replacing missing
values
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . N.
J. Cox
2/03 How can I replace missing values with previous or
following nonmissing values?
http://www.stata.com/support/faqs/data/missing.html
Nick
[email protected]
Olexandr O. Vasetsky
of course, remove the repeating "replace", my mistake
sort panelid timevar
by panelid: replace indep3 = indep3[_n-1] if indep3== ""
gsort panelid -timevar
by panelid: replace indep3 = indep3[_n-1] if indep3== ""
sort panelid timevar
Olexandr O. Vasetsky
to fill your indep3 dummy you can use something like this (might need to
"ssc install gsort" first):
sort panelid timevar
by panelid: replace replace indep3 = indep3[_n-1] if indep3== ""
gsort panelid -timevar
by panelid: replace replace indep3 = indep3[_n-1] if indep3== ""
sort panelid timevar
this program fills ALL the blanks for the indep3 - i.e those after the
existing observation (like in your example) as well as blanks BEFORE the
existing one, in case you have to fill it in both directions.
Max
> I am working on a panel dataset. Couple of variables varies overtime
> and across panels, others only vary across panels. Got some additonal
> data for time & panel varying variables, for new time periods. I would
> like append the new observations to the panel and then within each
> panel ensure the records for panel only variables are applied to them
> (i.e. no blanks).
>
> example:
>
> panelid timevar dep indep1 indep2
> indep3(dummy)
> 1 1/1/2000 321 899 12
> 1
> 1 2/1/2000 882 980 34
> 1
> add=> 1 3/1/2000 698 898 87 (here I
> want the above value to apply instead of a blank)
>
> 2 1/1/2000 338 768 90
*
* For searches and help try:
* http://www.stata.com/help.cgi?search
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
