Help- I seem to be having a perplexing problem.

I need to run a GEE model but it crashes in under 25 seconds due to an apparent lack of memory.

Why am I perplexed? Because: 1) this is a small subset of the data ( less than 200,000 observations) 2) when the same model has an additional nested variable added to the subject-effect line (i.e. repeated subject=MD(new variable) it doesn`t crash. Go figure.

Here's the log: 159 proc genmod data=two; 160 class MD ; 161 model x=a b c; 162 repeated subject=MD; 163 run;

NOTE: The scale parameter was estimated by maximum likelihood. NOTE: The SAS System stopped processing this step because of insufficient memory. NOTE: At least one W.D format was too small for the number to be printed. The decimal may be shifted by the "BEST" format. NOTE: The PROCEDURE GENMOD used 24.78 seconds.

I am using a powerful NT machine with at least 11 free GB of space and 128MB ram. How can one calculate the memory necessary for such models? Is there a formula or a rough guide?

If the un-nested model crashes and the more complicated nested one does not, do you think maybe I should be skeptical vis à vis the nested results?

Thanks in advance for your input,

Machelle Wilchesky Division of Clinical Epidemiology Royal Victoria Hospital, Montreal Que

and

Physical Health Unit Régie régional de la santé publique montréal-centre

Sent via Deja.com http://www.deja.com/ Before you buy.