The Effects of Visibility Bias on Time-Budget Estimates of Niche Breadth and Overlap

Abstract

Abstract Four time-budget estimation strategies are compared with respect to their sensitivity to two components of visibility bias in the observation process: discovery bias and loss bias. Monte Carlo simulations and a brief field study both indicate that visibility bias (particularly discovery bias) can substantially affect the results of time-budget studies. Estimators designed to curtail these biases performed best. Counting only initial contacts was least satisfactory. Bootstrap confidence intervals for niche overlap from the field study were so broad that overlap estimates seem nearly useless with very small sample sizes, such as the 93 observation series with 1,065 data points obtained here. Investigators who measure time or energy budgets in the field should take care to minimize sample biases, obtain adequate sample sizes, select analysis techniques appropriate for their sampling scheme, and confine inference to a scope compatible with the temporal and spatial scale of their study.