02-04-2015 04:29 PM
Hi I am a beginner in SAS. I have data regarding whether people withdraw or not based on issue age and tax type. From what I understand PROC LIFETEST will not take into account all the other covariates.
For this reason I used PROC LIFEREG with a Weibull distrbution. I have the fitted beta parameters as well as the scale, shape for the weibull but I am not sure how to get at the hazard function for each row of data? I am interested in getting the conditional probability of failure if you have survived up to that point. I tried
hazard function = scale*exp(-xbeta)*(shape)^(scale-1)
but the function does not seem to vary much with age. Any help?
02-06-2015 09:17 AM
If scale dominates shape, and shape is greater than 1, that function will be very flat, and look much like shape to the scale power for all values of age. You might consider rescaling age to a "smaller" number that still captures the full variability, but such that xbeta*age (which is what should be exponentiated, I think), and scale are of the same order of magnitude.