Floating Point Zero Divide in PHREG

[issac]

Hi folks;

I have a data set with repeated event times and want to explore the Fixed-Effect Partial Likelihood with the help of strata. I submit the following code and get the error "Floating Point Zero Divide".

proc phreg data =test  multipass nosummary ;

CLASS mar emp(ref='UNKNOWN') agents admsource ;

model rtime * rstatus (0) = mar emp agents admsource seq/ ties=efron;

strata  id / missing ;

run;

any helpful comments would be appreciative.

Thanks!

Issac

Message was edited by: issac shams I am working with SAS9.3.

[Rick_SAS]

Is that the only error you get? That sounds undesireable. Consider opening a track with SAS Technical Support: http://support.sas.com/ctx/supportform/createForm

[issac]

Yes it is. I found something related to this but not sure whether the problem is caused by ties handling or not?

http://support.sas.com/kb/7/140.html

[Rick_SAS · Accepted solution]

That article indicates that the "ties" problem was fixed in SAS 9. You say you are running SAS 9.3, so this is either a new problem (most likely) or your data exhibits a structure that isn't handled by the fix.

[issac]

As I know the data set contains a lot of class variables which can be handled by CLASS statement and there is nothing wrong with the it. So I may want to submit a SAS  Technical Report.