I could perform validate() on the orm() object, not the lrm() object. The latter gives me this message:
Error in predab.resample(fit, method = method, fit = lrmfit, measure = discrim, :
A training sample has a different number of intercepts (3)
than the original model fit (4).
You probably fit an ordinal model with sparse cells and a re-sample
did not select at least one observation for each value of Y.
Add the argument group=y where y is the response variable.
This will force balanced sampling on levels of y.
Hi Professor Harrel
I could perform validate() on the orm() object, not the lrm() object. The latter gives me this message:
Error in predab.resample(fit, method = method, fit = lrmfit, measure = discrim, : A training sample has a different number of intercepts (3) than the original model fit (4). You probably fit an ordinal model with sparse cells and a re-sample did not select at least one observation for each value of Y. Add the argument group=y where y is the response variable. This will force balanced sampling on levels of y.
Any possible help is very much appreciated.
All the best
Sandro