Effects of violation of assumptions for survival analysis methods in radiotelemetry studies

Two main methods of estimating survival from radiotelemetry studies are the Trent and Rongstad approach and the Kaplan-Meier approach. Although these 2 methods often are applied, the effects of violation of assumptions have n ot been completely studied. We present a very brief evaluation of all assum ptions and their violation. We also evaluated the assumption of equal censo ring rates for live and dead animals by simulation and found that large bia ses can result if it is violated. For the Trent-Rongstad method where there is an additional assumption of constant survival rates over time, our simu lations showed that violation of this assumption caused the largest bias wi th an early drop in survival rate as compared to other alternatives.