st: Re: RE: Re: Fill in panel observations

["Martin Weiss" <martin.weiss1@gmx.de>]

<>

That is the thing about posts with example data in them. You can answer the 
request based on the structure of the data that you observe in the post. 
That means you assume that the poster knows best and gave an accurate 
representative sample of his data.

Or you think ahead and anticipate that there must be some kind of  problem 
in there that the poster does not regard as a problem, e.g. missing values. 
But you run into a new problem then: We do not know what John wants to 
happen if there are missing values. Maybe he still wants the -egen, count()- 
solution even then?

I chose the former attitude to the problem, but I recommend that John pay 
close attention to Nick`s recommendation and implement them if he feels they 
do what he wants...


HTH
Martin
_______________________
----- Original Message ----- 
From: "Nick Cox" <n.j.cox@durham.ac.uk>
To: <statalist@hsphsun2.harvard.edu>
Sent: Sunday, March 29, 2009 6:02 PM
Subject: st: RE: Re: Fill in panel observations


> There will be a difference between the results
>
> bys iis: egen total_rater = count(rater)
>
> and
>
> bys iis : gen total_rater = _N
>
> if -rater- is ever missing; but otherwise the latter is more direct and
> more efficient.
>
> Incidentally, once you known that data are sorted on -iis-, the second
> command can just start -by iis:-.
>
> Nick
> n.j.cox@durham.ac.uk
>
> Martin Weiss
>
> bys iis: g rater=_n
> bys iis: egen total_rater=count(rater)
>
> John Antonakis
>
> > I have multiple observations on leaders (iis) by raters, like this:
> >
> > iis  sex  age   y  rater   total_rater
> > 1    0    43    5    1        5
> > 1    0    43    9    2        5
> > 1    0    43    3    3        5
> > 1    0    43    6    4        5
> > 1    0    43    6    5        5
> > 2    1    46    9    1        7
> > 2    1    46    7    2        7
> > 2    1    46    9    3        7
> > 2    1    46    6    4        7
> > 2    1    46    7    5        7
> > 2    1    46    2    6        7
> > 2    1    46    8    7        7
> >
> > At this time, the rater column and total_rater columns are blank. I
> > would like to fill them in such that:
> >
> > a. Under "Rater" I number each within the panel, starting with 1 and
> > going up to k raters (as I have done above).
> > b. The total number of k raters is included for each leader (as in the
>
> > total_rater column).
> >
> > Is there a quick way to do this?
>
> *
> *   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/

*
*   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/