Estimating Prey Preference by Predators: Uses of Various Indices, and a Proposal of Another Based on χ²

Abstract

A really useful prey-selection index should be statistically testable. A 2 × 2 contingency table formulation provides a simple way to test the significance of any degree of selection at any sample size. An index ranging from −1 to +1, and zero-valued for no selection, can be simply derived from the chi-square formulation. Some conflicting usages in biological statistics texts are discussed, as well as the usefulness and limitations of other commonly used indices.Key words: selection, index, chi-square, exact test, prey