Proc surveylogistic (marginal probabilites)

[gtmbini]

I want to calculate marginal probability for multinomial logistic regression. I have complex survey data and I am using PROC survey logistic to built the model. I was looking for LSMEANS but it assumes balanced population. I was wondering if there are ways to account for actual proportion of the data.

Please let me if you have better idea to present the multinomial result as probabilities rather than OR.

Thank you 

[SAS_Rob]

The OM= option on the LSMEANS statement will allow you to compute the LSMEANS based on a marginal weighting scheme that is unequal.  Check the documentation for the specific details.

https://go.documentation.sas.com/?cdcId=pgmsascdc&cdcVersion=9.4_3.4&docsetId=statug&docsetTarget=st...