CORRECTING NESTING-SUCCESS ESTIMATES FOR OBSERVER EFFECTS: MAXIMUM-LIKELIHOOD ESTIMATES OF DAILY SURVIVAL RATES WITH REDUCED BIAS

Abstract

Abstract We develop a statistical method that simultaneously estimates daily survival rate and observer effect. We used Monte Carlo simulation to (1) evaluate the performance of the model, (2) compare model performance with models that ignore observer effects, and (3) evaluate methods of choosing between competing models of survival. When observer effects were absent, all models produced unbiased estimates of daily survival rate. In the presence of observer effects, however, models that ignore these effects underestimated daily survival rate. In such cases, estimates of nesting success were strongly affected even when observer effects were relatively small. In contrast, estimates of daily survival rate and nesting success produced by the model that considers observer effects consistently had little bias. However, estimates of daily survival rate from this model were less precise than those from the simpler model. Objective criteria for choosing between competing models did not perform well with sample siz...