Trying to fit a model using the Surv() with a frailty term but get the following cryptic error message: ------------------------------------------------------------------------------ Error in frailty.brent(sqrt(x), y, lower = 0) : Ties for max(y), I surrender Calls: runplots ... coxpenal.fit -> eval -> eval -> <Anonymous> -> frailty.brent ------------------------------------------------------------------------------ When I remove the frailty term from the model it runs just fine. Does anyone know what this error message means? Thanks, M.
--- begin included message ----- T rying to fit a model using the Surv() with a frailty term but get the following cryptic error message: ------------------------------------------------------------------------------ Error in frailty.brent(sqrt(x), y, lower = 0) : Ties for max(y), I surrender Calls: runplots ... coxpenal.fit -> eval -> eval -> <Anonymous> -> frailty.brent ------------------------------------------------------------------------------ When I remove the frailty term from the model it runs just fine. Does anyone know what this error message means? Thanks, M. -------- end included message ----------- What it means is that the function maximization routine "under the covers" is completely befuddled. (I saw the "I give up" message fairly often in the earliest drafts of the code.) Solution: download the coxme package from CRAN -- superior in every way to frailty() terms. For one thing I made the maximizer much more intelligent: I now use optim :-). Terry Therneau