Can I force the direction of a hazards ratio?

[Ubai]

Hi everyone,

I am doing a survival analysis using proportional hazards' model with several predictive variables. Some of the estimated hazards ratio go in an implausible direction, which contradicts what we know from the literature. For example, smoking is protective !!

I was wondering if I can force SAS to estimate HR in one direction or in a certain range.

[Doc_Duke]

No.  The problem that you probably have is that the model has been overfit.  This typically happens when yo have too many predictor variables for the number of events.  See

https://en.wikipedia.org/wiki/One_in_ten_rule