Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: -foreach- and -forvalues-


From   "Katsuhide Isa" <CQB01114@nifty.ne.jp>
To   <statalist@hsphsun2.harvard.edu>
Subject   Re: st: -foreach- and -forvalues-
Date   Sun, 22 Dec 2002 00:02:15 +0900

To Uli and Kit
:
Thank you for your advice.
Sorry for my late reply, as it took some time to verify
the uploaded codes.


------------------------------------------------
(Uli wrote)

> forvalues i = 1997/2001 {
>      use "personnel_`i'"
>      sort id
>      gen year = `i'
>      save "using", replace
>      use personnel
>      sort id
>      capture drop  _merge
>      merge id using "using""
>      save personnel, replace
> }
> erase "using.dta"


> * Prepare using data
> forvalues i = 1997/2001 {
>      use "personnel_`i'"
>      sort id
>      gen year = `i'
>      save "using_`i', replace
> }
>
> * Load Master Data
> use personnel
> sort id
>
> * Merge Data
> forvalues i = 1997/2001 {
> merge id using "using_`i'"
>         drop _merge
>         erase using_`i'.dta        /* Clean up things */
> }
>
> save personnel_new, replace

I didn't know -forvalues- could handle files as well as
variables. Then, -foreach- is not required at all in this
case.
Of the two routines above, the latter one may be
better(with some modifications), as you suggested.

> Finally you might want to check out -mmerge- on SSC
> which makes merging much easier.

Thank you for your recommendation. I didn't know of
this module.


------------------------------------------------
(Kit wrote)

> Why do you want the time variables to be 'year1997', 'year1998' and so on?
> It would seem obvious that 'year1997' == 1997, or perhaps 1 if it is a
> dummy for year, but it would seem sufficient to have a single variable
> 'year' which contains the year of each chunk of data.

Maybe you are right.
I tried to create 'year1997', 'year1998' and so on simply
because I didn't know a better way.

> Further if these files are unit records -- e.g. one record per employee
> per year -- don't you want to append them, rather than merging them?
> Unless you have different variable names in the various files, merging
> them will not likely be what you want (and if you do, it is even less
likely
> to be what you want -- it will create a block-diagonal structure with
> mostly missing data). -append- would create a panel in 'personnel' with a
> t-indicator of year. Something like
>
> forv i=1997/2001 {
> use personnel_`i',clear
> gen year = `i'
> save personnel_`i',replace
> }
> use personnel,clear
> forv i=1997/2001 {
> append using personnel`i'
> }
> save personnel,replace

The structure of the data files at hand is as follows:

personnel_1997.dta  personnel_1998.dta ...  personnel_2001.dta
------------------------------------------------
         id   year             id  year                       id  year
  101818  1997      101818  1998               101818  2001
  102270  1997      102881  1998               103454  2001
  102881  1997      103454  1998               104043  2001
  103241  1997      103586  1998               104124  2001
  103373  1997      103799  1998               104132  2001
  103438  1997      104043  1998               104710  2001
  103454  1997      104124  1998               104868  2001
  103586  1997      104132  1998               105210  2001
  103799  1997      104281  1998               105635  2001
  103888  1997      104523  1998               105708  2001
  ...

Namely, there are attrition and/or entry of employees
during the panel period, though most of them survive
from start to end.
So, I thought using -merge- with each file matched by id
variable would be a better choice.
I'd appreciate additional advice if I am wrong.

> Note, btw, that
>
>    capture confirm variable _merge
>    if _rc {
>    continue
>    }
>    else {
>    drop _merge
>    }
>
> can be replaced by
>
> capture drop _merge

Oh, I didn't know such a simple code could handle errors
in Stata.
Error handling of this kind would be more complicated in
other script languages

K.I.


*
*   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/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index