However feeble you think your code may be (& I have certainly authored some
very primitive stuff & still do), it would
allow others to better assist you were you to share it. With the data
clean up you have already performed, it certainly
sounds like you are able to do some things in Stata. Nonetheless, to get
data in long format, you'd want to check
out the reshape command.
Without knowing more, it's hard to see how & what looping your project
might actually require. If you plan on
looping through variables on a given observation, then you might not want
things in long format. Or you might
want to loop through the variables first & then reshape. At least show us
what the end result should look like --
formulating that is often very helpful in getting yourself to the desired
solution.
I look forward to hearing more.
Eric G. Wruck
Original Message:
-----------------
From: tony fong [email protected]
Date: Sun, 13 Nov 2005 12:57:37 -0800
To: [email protected]
Subject: st: Re: assistance writing a loop to create a panel dataset
Hi all,
Nick mentioned that I should show the code I've been working on to change
my
dataset into panel data, but I haven't been able to put anything together
worth showing as I'm new to writing code (so I'm still at square one).
Below I also try to repost the example of what my data currrently looks
like, as it was kind of tough to read in the last email. If anyone could
assist me it'd be greatly appreciated.
Thanks!
id sch1 sch2 start1 end1 start2 end2
1 73 . 199308 200000 . .
2 46 . 199307 199506 . .
3 12 16 199308 199505 199508 199606
4 94 58 199408 199412 199501 199606
5 24 . 199608 199706 . .
6 84 43 199308 199605 199608 199611
7 16 . 199508 200000 . .
8 59 87 199309 199406 199609 200000
where sch1 and sch2 represent the school ID that the teacher was teaching
at
for the respective teaching spells 1 and 2, and start1-end2 represent the
start and end dates of spells 1 and 2, where the first four digits is the
year and the last two digits is the month of that year (actually my data
has
five possible start and end date spells, but it got too messy in pasting
the
info above). If the end date is 200000, then the teacher remained at that
school until the data was right-censored in 1998. I've cleaned the data so
that all of the spells are in order and none of them overlap (although
there
are gaps of time between some teaching spells). What I'm trying to do is
create a panel data set in long format, with one observation for each year
from 1992 until 1997 for every individual. It would give the school that
the
individual taught at for each year (where for the year 1995, for instance,
it would correspond to the school number that the teacher taught at for any
of the months from 199507 through 199606, inclusive), or it'd be a missing
value if the teacher did not teach during that year since some teachers
started later and many teachers quit after only a year or two. I've tried
to
create a set of loops that loops through each consecutive non-missing spell
to fill in the variables school92-school97 (which represent the school
taught at for each year) based on the start and end dates of each spell,
but
I've been unable to do it. Also, some teachers teach at one school for half
the year and another school for the rest of the year -- in this case the
first school should count for that particular year. If anyone can help me
out with the code that would accomplish this task it would be very much
appreciated.
Thanks!
Tony
*
* 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/
--------------------------------------------------------------------
mail2web - Check your email from the web at
http://mail2web.com/ .
*
* 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/
