Dear listers,
I had a problem while I was trying to replicate an old
trick as described in Aitkin and Clayton (Appl.
Statist. 1980 29:156-163).
These authors described how one can fit a Weibull
parametric survival model in GLIM using an equivalent
Poisson model. Application of this idea was
straightforward in STATA and the procedure gave the
same coefficients as the "streg,dist(weib)" command
but the S.E.s were underestimated as expected because
this procedure does not allow for the additional
estimation of the shape parameter. However the authors
describe a technique based on the calculation of the
observed information matrix to adjust the var-cov
matrix of the estimates. I wrote a few lines of code
to implement this method but the adjusted variances
were still smaller than those obtained by the streg
command (but larger than the unadjusted ones).
Any explanation for this discrepancy assuming that my
code is correct? Any thoughts are wellcome.
Nikos Pantazis
__________________________________
Do you Yahoo!?
The New Yahoo! Search - Faster. Easier. Bingo.
http://search.yahoo.com
*
* 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/
