Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

RE: st: discrete-time survival analysis

From   "Svend Juul" <[email protected]>
To   <[email protected]>
Subject   RE: st: discrete-time survival analysis
Date   Thu, 2 Jun 2005 15:02:42 +0200

Luciana wrote:
I am analyzing the marriage event using a two-wave survey (2001 & 
2003). The dependent variable =0 if the respondent is unmarried and 1 
otherwise. The purpose is to analyze whether a program implemented 
after the 2001 survey delayed marriage among the respondents. 

I was advised to use discrete-time survival analysis using logistic 
regression in STATA, with 2 observations per respondent. But I have 
two unanswered questions:

(1) I was told to include those who were already married in 2001. Does 
this make sense?
(2) Also, all the studies I have read using discrete-time survival 
analysis have more than 2 observations per respondent. Is it a problem 
that I only have 2?

This is a bit unclear to me. I assume, but I may be wrong:
- that the same persons were asked in 2001 and 2003
- that if a person became married between the first and second survey,
you don't know when (otherwise you would not need a discrete-time
- that the intervention took place some time between 2001 and 2003.
- that "unmarried" means never married, so you can't change stats from
married to unmarried (divorce).

Obviously those who were already married in 2001 are not at risk to
become married. But I don't see how you could produce evidence on the
effect of an intervention if my assumptions are right. You must produce
some more information.



Svend Juul
Institut for Folkesundhed, Afdeling for Epidemiologi
(Institute of Public Health, Department of Epidemiology)
Vennelyst Boulevard 6
DK-8000  Aarhus C, Denmark
Phone: +45 8942 6090
Home:  +45 8693 7796
Email: [email protected]

*   For searches and help try:

© Copyright 1996–2024 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index