[S] using glim for data from Hakulinen's relative survival package

Zhaorong Jiao (z.jiao@qut.edu.au)
Tue, 27 Oct 1998 20:04:08 +1000 (EST)


Hi, I posted a similar message before. As someone pointed out to me that I
did not put a subject to the message. It is quite right. It just happened
that I forgot doing so last time. I apologise.
What I was tring to say in my last message is that: does anyone have some
splus codes for analysing the data from Hakulinen's relative survival
package in the way described in Hakulinen's 1987 paper ? Or perhaps the
splus function glim can do the job ?
a data set from Hakulinen's relative survival package contains the number
of death, the number of patients at risk and the expected survial rate for
each of the combinations between follow-up intervals and levels of
predictors. The data set also contains the coding for predictors.
Hakulinen has developed four macros in the GLIM package for analysing the
data set to explore the relationship between relative survival ratio and
predictors. The relative survival ratio is defined as observed survival
rate devided by expected survival rate. The model proposed for relative
survival ratio is given by

log(-log(observed survival/expected survival))=linear combination of
predictor
with a binomial error. I do not know the GLIM package. So I used the splus
function glim to fit the data using the expected survival rate as the
weight(link function was set to "loglog"). The result looked OK but I am
not sure if I have done the right thing. Can anyone help me with this
problem ? Thak you.

:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::_--_|\::::
:: Zhaorong(Jo) JIAO Ph: +61 7 3864 5267 / QUT::
:: School of Mathematics, GP. QUT, Fx: +61 7 3864 2310 \_.--._/ ::
:: PO Box 2434, Brisbane Q 4001, Australia Em: z.jiao@fsc.qut.edu.au V ::
:::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::

-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news