re:st: Creating variable based on time reference

[Christopher Baum <kit.baum@bc.edu>]

<>
Matthew said

I have a data set with 3 variables -- "id", "datesubmitted", and "group" 
-- and 5 observations.

id    datesubmitted    group
1.    1oct2011    1
2.    1aug2011    2
3.    2aug2011    2
4.    5aug2011    2
5.    9oct2011    1

I would like to create a new variable called "responsetime" which 
reflect the amount of time elapsed (categorically) between the first 
response and each later observation within a group. For example, if an 
observation occurred within two days of the first response from that 
group, it would be a 1; between two and five days, it would be a 2; six 
and greater would be a 3. In this case, IDs 1 and 5 would have a 
"responsetime" of 1 and 3, respectively. IDs 2, 3, and 4 would have 
"responsetime" of 1, 1, and 2, respectively.

I'd like to do this across thousands of observations and dozens of 
groups. Is there a simple way to do this that the trusty Google didn't 
turn up?


g datesubmitted = date(datesub,"MDY")
bys group: g delta = datesubmitted - datesubmitted[1]

. l

     +--------------------------------------+
     |   datesub   group   datesu~d   delta |
     |--------------------------------------|
  1. | 10/1/2011       1      18901       0 |
  2. | 10/9/2011       1      18909       8 |
  3. |  8/1/2011       2      18840       0 |
  4. |  8/2/2011       2      18841       1 |
  5. |  8/5/2011       2      18844       4 |
     +--------------------------------------+




Kit Baum   |   Boston College Economics & DIW Berlin   |   http://ideas.repec.org/e/pba1.html
                             An Introduction to Stata Programming  |   http://www.stata-press.com/books/isp.html
  An Introduction to Modern Econometrics Using Stata  |   http://www.stata-press.com/books/imeus.html


*
*   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/