Prev: Populating variables
Next: ERROR: Insufficient page size to print frequency table.
From: saslog on 5 Oct 2009 10:30
I am trying to using proc qlim to carry out a logistic regression. I
want to compute the marginal effects and also the predicted
probabilities at the sample means. For this I append to the dataset on
which the model is computed, extra observations with means of the
explanatory variables, but missing value of the dependent variable, so
that model fitting is not affected. However, SAS does not seem to
compute the predicted probabilities for these observations. If I use
proc logistic, I am able to get predicted probabilities for
observations not used in the model fitting. Am I doing something
wrong? Here is my code:

data me_mean;
input var_a;
datalines;
0.5;
run;

data me_mean;
set modeldata me_mean;
run;

proc qlim data=me_mean;
model y=var_a /discrete (d=logit);
output out=outme1 marginal;
run;
quit;

If I use proc logistic for the same, the out command does give me the
predicted probabilities also evaluated for var_a=0.5 but not qlim !

Thank you for your help.
 | 
Pages: 1
Prev: Populating variables
Next: ERROR: Insufficient page size to print frequency table.