st: stset for conditional risk set (multiple failure) models

[Layna Mosley <lmosley@email.unc.edu>]

I'm new to duration models generally and to repeated events models 
specifically. I'm having difficulty with getting the correct syntax for 
the stset command, so that I can then run a conditional risk set (gap 
time) model (as in Cleeves 1999).

My data are panel data (country-year format), with repeated failures. I 
have generated a sequence/stratification variable (to be used later in 
the analysis), as well as variables for overall time (by country), and 
conditional time (resetting the time count with each failure within the 
country). I also have -- although I perhaps don't need -- an "enter" 
variable, set to one for the first observation by country, and then 
reset to one in any period immediately after a failure.

My question is how to use this information with stset. The Cleeves 
article suggests
  
   stset time,fail(status) exit(time .) enter(time0).

For my variables and data, I think this would mean

   stset(duration), failure(turkredenom) id(countrynumber) exit(time .) 
enter (time0)    
This model generates a probable error, though, and I'm guessing it's 
because the starting interval ends up as duration rather than as zero. 
It's also unclear to me how/whether to use the time0 option; the Stata 
documentation suggests it's "rarely used." At this point, I've just 
generated time0=0, but I'm sure that's wrong.  And I'm not quite sure 
how to put together my variables into the stset commands. I'd appreciate 
any advice...these sorts of models don't seem widely used/documented.

Thanks!

Layna


.     stset(duration), failure(turkredenom) id(countrynumber) exit(time 
.) enter (time0)

               id:  countrynumber
    failure event:  turkredenom != 0 & turkredenom < .
obs. time interval:  (duration[_n-1], duration]
enter on or after:  time time0
exit on or before:  time .

------------------------------------------------------------------------------
    5861  total obs.
     719  multiple records at same instant                     PROBABLE 
ERROR
          (duration[_n-1]==duration)
------------------------------------------------------------------------------
    5142  obs. remaining, representing
     160  subjects
      18  failures in single failure-per-subject data
    5142  total analysis time at risk, at risk from t =         0
                            earliest observed entry t =         0
                                 last observed exit t =        44



*
*   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/