Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: discrete hazard modells: irregular time intervals


From   Robert A Yaffee <bob.yaffee@nyu.edu>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: discrete hazard modells: irregular time intervals
Date   Thu, 27 May 2010 02:14:06 -0400

Oliver,
 
  If you would like to read chapter five in Applied Longitudinal Data
Analysis by Judith Singer and John Willett (2003)Oxford University Press,
they discuss several approaches to dealing with irregularly spaced waves
in longitudinal analysis.  According to Singer and Willett, it is possible
to center age or age-group as a functional equivalent of a time or wave
designation.  If you introduce more structure to the time effect than is
warranted, the model they say will fit less well.  Another approach is to
use the time of interview as the temporal indicator of the wave.  You
may want to transform the time variable to permit it to be constant after
which you might re-transform the fitted model to obtain the proper 
interpretation of the effects, if the model is not too unbalanced by
these effects to begin with.  You might analyze in terms of rates of change rather than levels but transform back to levels for your interpretation.
 
     Disentangling, time, period, and cohort effects is often a challenge
in longitudinal data. With seriously unbalanced data, Singer and Willet contend that estimation of the missing may values may prove difficult if too high a percentage of the respondents have incomplete non-ignorable data. With such data, residual variances may become difficult to properly estimate as well.
Under such conditions, computer algorithms may have difficulty with boundary constraints or convergence.    When this occurs, you are advised by the authors to remove the offending variables after careful deliberation rather than let the computer try to decide for you which variables to drop. Sometimes, the choice of better starting values helps. If there is an option for selecting random starting values, you might go for this approach to see which model converges.
Otherwise, you have to ask yourself whether you really have enough degrees of freedom to test the complexity of your model.   Of course, you could try other algorithms or more robust estimators of variances if not nonparametric or semiparametric estimators.  You should check out this chapter in their excellent book.
       Regards,
       Robert
     
Robert A. Yaffee, Ph.D.
Research Professor
Silver School of Social Work
New York University

Biosketch: http://homepages.nyu.edu/~ray1/Biosketch2009.pdf

CV:  http://homepages.nyu.edu/~ray1/vita.pdf

----- Original Message -----
From: Robert A Yaffee <bob.yaffee@nyu.edu>
Date: Thursday, May 27, 2010 0:26 am
Subject: Re: st: discrete hazard modells: irregular time intervals
To: statalist@hsphsun2.harvard.edu


> Oliver,
>   Are your intervals small enough to aggregate them into 
> larger but temporally equal intervals?  That way you can
> still do the discrete time logistic regression using the hazard
> rate in the odds ratio.
>   You might consider a finite mixture model where one of the
> components dealt with the duration of the risk set and the other
> dealt with the hazard rate within it.  You could alternatively 
> weight the risk by the exposure time to compensate for the
> irregular duration of the periodization if you could find a 
> relationship between the rate and duration that was strong enough.
>    You would have to make the adjustments for the interval
> censoring of course.
>    -  Robert
> 
> Robert A. Yaffee, Ph.D.
> Research Professor
> Silver School of Social Work
> New York University
> 
> Biosketch: http://homepages.nyu.edu/~ray1/Biosketch2009.pdf
> 
> CV:  http://homepages.nyu.edu/~ray1/vita.pdf
> 
> ----- Original Message -----
> From: Oliver Eger <oliver.eger@yahoo.de>
> Date: Wednesday, May 26, 2010 8:14 am
> Subject: st: discrete hazard modells: irregular time intervals
> To: statalist@hsphsun2.harvard.edu
> 
> 
> > Dear Stata users,
> > 
> > I ?am new to Stata and to the newsgroup. I would kindly ask for advice,
> > concerning my work on survival analysis.
> > 
> > I collected company data. My data are interval censored and interval
> > truncated. The intervals are of irregular period in calendar time. 
> > 
> > Are there any methods to deal with this irregularities or at least to
> > estimate their influence on my survival regressions?
> > 
> > Best regards,
> > Oliver
> > 
> > 
> > ================================================
> > Oliver Eger
> > Lehrstuhl für VWL - Innovationsökonomik
> > Universität Augsburg
> > Universität Hohenheim
> > Prof. Dr. Horst Hanusch - Prof. Dr. Andreas Pyka
> > ================================================
> > 
> > 
> >  
> >  
> > 
> > 
> > 
> > *
> > *   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/

*
*   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/


© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   Site index