Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

st: R: a person-period multi-episode event history file


From   "Carlo Lazzaro" <carlo.lazzaro@tin.it>
To   <statalist@hsphsun2.harvard.edu>
Subject   st: R: a person-period multi-episode event history file
Date   Tue, 6 Oct 2009 19:08:46 +0200

Katya wrote:


<Basically, I have 1513respondents and each one has been followed for 60
months (which are not identical for all respondents ? for example, one is
followed from Nov 01 to Nov 06 and another one from Jun 02 to Jun 07, and so
on).>

This would seem a minor matter of concern, since I suppose that is the
observation span of time that matters. Hence, you can set t0 when the
subject comes to your attention unless enrolling subjects at different times
may influence, for whatever reason , the result of your analysis.

<What I need is to create a 0/1 variable for each one of those months
denoting whether or not a certain event has taken place for that
person. It is easy when it is an event which occurs only once (e.g.,
divorce of biological parents). It gets more complicated however, when
I have events which not only can occur multiple times but also have
different durations.>

Are you sure you cannot switch to Stata built-in Survival Analysis
procedures?

Otherwise, you can create an ad hoc variable and define value labels for
each event you are interested in (please, type - help label - in Stata)

label define event 1 "unemployment"
label define event 2 "on the job training"
label define event 3 "on holidays"

and so on and so forth

Then, you can:

count if event==1 & id==1 &  month=="Jan_01"

<What I need to do now is create a ?dating status?
variable per month which has the value of 1 if the adolescent was
dating in that month and a value of 0 if they weren?t.>

This is a very temptative answer:
g dating_index=1 if Start rel_n !="." $	End rel_n=="."

Sorry I cannot be more helpful.
Kind Regards,
Carlo
-----Messaggio originale-----
Da: owner-statalist@hsphsun2.harvard.edu
[mailto:owner-statalist@hsphsun2.harvard.edu] Per conto di K.O.Ivanova
Inviato: lunedì 5 ottobre 2009 17.02
A: statalist@hsphsun2.harvard.edu
Oggetto: st: a person-period multi-episode event history file

Hello, all.

I am struggling with the following problem: I am trying to create a
person-period multi-episode event history file. Basically, I have 1513
respondents and each one has been followed for 60 months (which are
not identical for all respondents ? for example, one is followed from
Nov 01 to Nov 06 and another one from Jun 02 to Jun 07, and so on).

What I need is to create a 0/1 variable for each one of those months
denoting whether or not a certain event has taken place for that
person. It is easy when it is an event which occurs only once (e.g.,
divorce of biological parents). It gets more complicated however, when
I have events which not only can occur multiple times but also have
different durations.

In my case, that problem comes from the ?dating status of the
adolescent? variable. What I have right now are two variables per
dating episode, one denoting start of dating episode and the other
one, end of dating episode. I have also created 60 variables per
respondent denoting the end of each ?individual-specific? month of
observation. What I need to do now is create a ?dating status?
variable per month which has the value of 1 if the adolescent was
dating in that month and a value of 0 if they weren?t.

Basically, I need to get from something like this:

Id 	Start rel 1	End rel 1	Start rel 2	End rel 2
1	Apr 01          Jul 01          Sept 01         Dec 01
2	May 02	        still going on		


To something like this:


Id    Calendar mo    # of mo of obs     Dating status    Dating index
          of obs

1	Jan 01	             1	               0          0
1	Feb 01	             2                 0            0
1	Mar 01	             3	               0	        0
1	Apr 01	             4	               1	        1
1	May 01	             5	               1	        1
1	Jun 01	             6	               1	        1
1	Jul 01	             7	               1          1
1	Aug 01	             8	               0	        0
1	Sep 01	             9	               1	        2
1	Oct 01	             10	               1	        2
1	Nov 01	             11	               1	        2
1	Dec 01	             12	               1	        2
2	Mar 02	             1	               0	        0
2	Apr 02	             2	               0	        0
2	May 02	             3	               1	        1
2	Jun 02	             4	               1	        1
2	Jul 02                   5	                1	         1
2	Aug 02	             6	               1	        1
2	Sep 02	             7	               1	        1
2	Oct 02	             8	               1	        1
2	Nov 02	             9	               1	        1
2	Dec 02	             10	               1	        1
2	Jan 02	             11	               1	        1
2	Feb 02	             12	               1	        1


To answer a possible question ? yes, I do need to maximally expand the
data. I cannot stop the clock at first change in dating status. I want
to have all the information from start of observation to interview
date.


Can anyone help me out with the syntax? Thank you!


Katya
*
*   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/



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