st: stset multiple-spell(recurrent) delayed entry

[Rodrigo Alvarez <rbpalvarez@gmail.com>]

stset multiple-failure (recurrent) left and right-truncation with
delayed entry and time-varying covariates

Dear Statalisters,

I have a problem in the implementation of stset for survival analysis
and wondered if anyone could help me.

(My question is also how to adjust the data to stset so that Stata
understands that the "observation window" is not the same for all id,
I'm not sure if I'm doing this
in the right way)

Specifically, I am studying the entry and exit of firms in export
markets. The entire period that I notice is from 1998 to 2010, but I
do not
necessarily have information on all firms during this period, i.e. I
have firms that are observed between 2000 and 2007, other
between 2001 and 2005 etc.

The event (failure) is the firm exit the export market when it was
exporting in the previous year.

Basically, I have to stset the data in Stata and I do not know the
specific command or routine related to the data structure that I have,
which is like an unbalanced panel in terms of Panel Data

I tried:
stset time1 , id(id) failure(failure) origin(dummy_exp==1)
time0(time0) exit(time1)

The firm 1, e.g.,was not at risk until 2006, when it became an
exporter, and exported for 1 year (survival time)

The output is as follows:

id	 begin_obs	 time0 time1 dummy_exp failure	_st	_d	_origin	_t	_t0
1	1	1998	        1998	1999	0	        0	0		2007		
1	2	1998	        1999	2000	0	        0	0		2007		
1	3	1998	        2000	2001	0	        0	0		2007		
1	4	1998	        2001	2002	0	        0	0		2007		
1	5	1998	        2002	2003	0	        0	0		2007		
1	6	1998	        2003	2004	0	        0	0		2007		
1	7	1998	        2004	2005	0	        0	0		2007		
1	8	1998	        2005	2006	0	        0	0		2007		
1	9	1998	        2006	2007	1		0	0		2007		
1	10	1998	        2007	2008	0		1	0		2007		
1	11	1998	        2008	2009	0		0	0		2007		
1	12	1998	        2009	2010	0		0	0		2007		
1	13	1998	        2010	2011	0		0	0		2007		
2	1	1998	        1998	1999	1		0	0		1999		
2	2	1998	        1999	2000	1		0	0		1999		
2	3	1998	        2000	2001	1		0	0		1999		
2	4	1998	        2001	2002	1		0	0		1999		
2	5	1998	        2002	2003	1		0	0		1999		
2	6	1998	        2003	2004	1		0	0		1999		
2	7	1998	        2004	2005	1		0	0		1999		
2	8	1998	        2005	2006	1		0	0		1999		
2	9	1998	        2006	2007	1		0	0		1999		
2	10	1998	        2007	2008	1		0	0		1999		
2	11	1998	        2008	2009	0		1	0		1999		
2	12	1998	        2009	2010	0		0	0		1999		
2	13	1998	        2010	2011	0		0	0		1999		
3	1	2000	        2000	2001	0		0	0				
3	2	2000	        2001	2002	0		0	0				
3	3	2000	        2003	2004	0		0	0				
3	4	2000	        2004	2005	0		0	0				
3	5	2000	        2005	2006	0		0	0				
3	6	2000	        2006	2007	0		0	0				
3	7	2000	        2007	2008	0		0	0				
3	8	2000	        2008	2009	0		0	0				
3	9	2000	        2009	2010	0		0	0				
3	10	2000	        2010	2011	0		0	0				
4	1	2004	        2004	2005	1		0	0		2005		
4	2	2004	        2005	2006	1		0	0		2005		
4	3	2004	        2006	2007	1		0	0		2005		
4	4	2004	        2007	2008	1		0	0		2005		
4	5	2004	        2008	2009	0		1	0		2005		
.
.
.

And also tried:

stset time1 , id(id) failure(failure) time0(time0) origin(time0==1998)

and got:

id	begin_obs	  time0 time1 dummy_exp failure _st	_d	_t	_t0
1	1	1998		1998	1999		0	0	0			
1	2	1998		1999	2000		0	0	1	0	1	0
1	3	1998		2000	2001		0	0	1	0	2	1
1	4	1998		2001	2002		0	0	1	0	3	2
1	5	1998		2002	2003		0	0	1	0	4	3
1	6	1998		2003	2004		0	0	1	0	5	4
1	7	1998		2004	2005		0	0	1	0	6	5
1	8	1998		2005	2006		0	0	1	0	7	6
1	9	1998		2006	2007		1	0	1	0	8	7
1	10	1998		2007	2008		0	1	1	1	9	8
1	11	1998		2008	2009		0	0	0			
1	12	1998		2009	2010		0	0	0			
1	13	1998		2010	2011		0	0	0			
2	1	1998		1998	1999		1	0	0			
2	2	1998		1999	2000		1	0	1	0	1	0
2	3	1998		2000	2001		1	0	1	0	2	1
2	4	1998		2001	2002		1	0	1	0	3	2
2	5	1998		2002	2003		1	0	1	0	4	3
2	6	1998		2003	2004		1	0	1	0	5	4
2	7	1998		2004	2005		1	0	1	0	6	5
2	8	1998		2005	2006		1	0	1	0	7	6
2	9	1998		2006	2007		1	0	1	0	8	7
2	10	1998		2007	2008		1	0	1	0	9	8
2	11	1998		2008	2009		0	1	1	1	10	9
2	12	1998		2009	2010		0	0	0			
2	13	1998		2010	2011		0	0	0			
3	1	2000		2000	2001		0	0	0			
3	2	2000		2001	2002		0	0	0			
3	3	2000		2003	2004		0	0	0			
3	4	2000		2004	2005		0	0	0			
3	5	2000		2005	2006		0	0	0			
3	6	2000		2006	2007		0	0	0			
3	7	2000		2007	2008		0	0	0			
3	8	2000		2008	2009		0	0	0			
3	9	2000		2009	2010		0	0	0			
3	10	2000		2010	2011		0	0	0			
4	1	2004		2004	2005		1	0	0			
4	2	2004		2005	2006		1	0	0			
4	3	2004		2006	2007		1	0	0			
4	4	2004		2007	2008		1	0	0			
4	5	2004		2008	2009		0	1	0			
.
.
.
observing the variables _st (obs can be included in the study) , _d
(outcome at the end of that spell) , _t, _t0 (time span tha obs lived)
we can see that is not right...

id 1, e.g., is being treated as if it lasts for 9 years when his spell
lasted only one year!

any suggestions will be appreciated! papers, routines, blogs...!

Tks anyway

Rodrigo Alvarez
Economics - University of Sao Paulo - Brazil

*
*   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/