Hi - In biometric applications, there has been a lot of recent work on
joint modeling of survival and a continuous longitudinal response, where
the latter contains random effects and/or measurement error. For
example, see the Wulfsohn-Tsiatis approach. WIth all the sophisticated
econometric instrumental variable, etc. routines in Stata, I wonder if
there is one already there that essentially does the same thing as
Wulfsohn-Tsiatis, but is hiding behind some econometric name or
terminology.
In rudimentiary terms, my understanding of the Wulfsohn Tsiatis model is
that of two variables, say T and X , where T given X and other
covariates follows a Cox survival model, and that X is measured
longitudinally with sporadic repeated measurements for each of a number
of subjects. However we observe not X, but X + a random subject effect +
a random measurement error, where the errors have normal distributions.
The Wulfsohn-Tsiatis estimation methdology then uses an EM approach to
get maximum likehood estimates of the parameters.
Thanks
Al Feiveson
*
* 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/
