I am a first-time poster in the list. I apologize if my questions turns out
to be too naive.
I'm interested in an event-history analysis (survival analysis) of a process
with at least three possible destinations. Quite unfortunately, I need to
use a discrete time model, because I just have interval-censored data
(European Community Household Panel).
Having read the lessons Stephen Jenkins generously offer in the ISER
web-page (University of Essex), I have learnt about the possibility of using
"discrete time proportional hazards (cloglog) model applied to interval
censored data". Episode-splitting and reorganizing the data seems then
necessary, with the expand command. Yet, the structure of my data does not
resemble the kind of data frequently used as an example; it's panel data.
I have information for each individual by wave. The problem is that same
waves are missing. Here I'll show an example
After creating a counter variable with the commands
bysort pid: ge time=_n
...I obtain the following results for some individuals.
pid wave time
1814103 1 1
1814103 2 2
1814103 3 3
1814103 6 4
1814103 7 5
As you see, there are some individuals for whom the time variable jumps from
3 to 4, when actually more than one year has passed.
How could I deal with this missing waves? how do they affect my clock?. Is
it a matter of modifying the time variable so that it captures the
additional time that has passed when some waves are missing?
This is my very simple question
Many thanks in advance for your patience and your help.
Luis Ortiz
*
* 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/
