Yes ... View more

No, you do not need to recode or change anything. It is just a note letting you know that if you chose to use different dummy variables you would end up with a different set of estimates, even though the fit would be the same. ... View more

I assume when you say you have the predicted values these are not predicted probabilities but some kind of classification based on the predicted probability.  If it is actually the predicted probability then you should be able to compute the residual in a data step.   If my assumption is correct then you could put the parameter estimates into a data set that is in the correct format for use as input into Proc LOGISTIC with the INEST= option and MAXITER=0.  You could then use the OUTPUT statement to recreate the predicted probabilities/residuals.    You could also use one of the methods here as well http://support.sas.com/kb/33/307.html       ... View more

You can basically ignore the message and.  It is a result of using the CLASS statement and the necessity of creating a series of dummy variables.  SURVEYREG over-parameterizes the model such that it sets the last level to zero, essentially making it a reference category, and then estimates the other parameters as compared to the referent.   It is more thoroughly explained in this usage note http://support.sas.com/kb/22585   Although that refers specifically to Proc GLM, what is said holds for SURVEYREG as well. ... View more

Unfortunately there is nothing in SAS to do this directly.  If you would be willing to fit a Poisson instead, using the SCALE= option then this might be of some help.     https://support.sas.com/rnd/app/stat/examples/SurveyPoisson/surveypoisson.htm    ... View more

To do a paired t-test you could simply create a difference variable and then use Proc SURVEYMEANS on that variable ... View more

There is no UNITS statement per se, but you can use the unit of change of the coefficient on the ESTIMATE statement.   estimate "X1-5 unit change" X1 5 / exp; ... View more

I believe that message is essentially saying that none of the replicate models converged.  I suspect you probably have some sort of separation in your model, leading to non-covergence.  Most likely the interaction is perfectly predicting the response--you could try running a Proc FREQ with  tables R_SMKCUR*R_RaceEthnicities*R_DSTRS30;   to diagnose it, or remove the interaction from the model and see if the problem goes away.     ... View more