LSMEANS in Logistic vs Genmod

[tka726]

Hello!

I am trying to calculate adjusted proportions for a binary outcome by quintile. I'm wondering why I get different results from LSMEANS in GENMOD & LOGISTIC. Both models are using the same reference groups (last quintile and non-male; age is continuous). If I remove the confounding variables (male and age) then the means from LSMEANS are the same. What am I missing here??

PROC GENMOD DATA=origs DESCENDING;
class quint;
model MI = quint male age;
lsmeans quint;
run;

PROC LOGISTIC DATA=origs descending;
CLASS quint / param=glm;
model MI= quint male age;
lsmeans quint / e ilink ;
run;

[PaigeMiller]

Oh come on, show us the output.

[tka726]

Sorry, I didn't realize that the output didn't attach.

The results for 'Estimate' in the least square means output section of Genmod are not the same as the 'Mean' in the least square means output of Logistic.

[PaigeMiller]

Sorry, but we probably ought to see screen captures of the output, pasted right into your reply. Most of us will not download Microsoft Office documents as they are a security threat.

Is it possible the LSMEANS are different because you use ILINK in one procedure but not the other?