ABSTRACT
An important outcome measured in ecological and environmental studies involves data on time to some event. Outcomes such as survival times of organisms or failure times of biological systems after exposure to an environmental insult generally exhibit statistical characteristics for which standard ANOVA-type analyses are inappropriate, even after some transformation of the observed data. The second class of survival regression models is the proportional hazards model, where the hazard function associated with a set of predictor variables is modeled as a multiple of some baseline hazard. The survfit function calculates the product-limit estimator for a survival object created by Surv for only one group, as denoted by the 1 on the right of the tilde. The simplest way of evaluating this question is via a graphical display: plot some transformation of estimated survival versus some transformation of time. If this pattern is roughly linear, then the proposed model is reasonably consistent with the data.
