Re: st: [ST] question: anything to do if "origin" is interval censored?

[Steven Samuels <sjsamuels@gmail.com>]

--

I can think of only one approach: Determine for a random sample of  
subjects (possibly by asking them) approximately when they received  
the invitation; gather enough information on all subjects to develop a  
predictive model for this; and apply MI techniques.  A Google search  
for '"interval censoring" "multiple imputation"' could be rewarding.

Steve

Steven J. Samuels
sjsamuels@gmail.com
18 Cantine's Island
Saugerties NY 12477
USA
Voice: 845-246-0774
Fax:    206-202-4783
On Oct 25, 2010, at 10:12 PM, László Sándor wrote:

Hi all,

I tried to check the following and have a feeling what to expect, I
still ask for your quick expert advice, which will be greatly
appreciated:

I am running an experiment where a subject becomes "at risk" once she
reads our invitation, but we only know when we sent her invitation and
the time she agreed to do what we asked (which clearly does not happen
upon receipt immediately). Is there anything I can do in this case for
Stata to be "robust" to this uncertainty?

I checked left censoring and truncation in the manuals, and I am also
uncertain how any method could deal with this kind of uncertainty in
the event time (also, uncensored "enter" time precedes censored
"origin"). Yet you might know something I missed, and I'm happy to
learn the prudent way to set things up (-stset-) in this case.

Thank you!

Laszlo


László Sándor
PhD candidate
Department of Economics
Harvard University

*
*   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/


*
*   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/