Comparison of Some Statistical Techniques for Analysis of Resource Selection

Abstract

Four statistical methods for analysis of habitat selection were compared using over 50,000 computer simulations of field data. Comparisons were based on Type I and Type II error rates, which varied among methods. The methods of habitat analysis gave contradictory results when applied to a given data set. Studies with few observations from few animals should be avoided because of unacceptable Type II error rates, and the number of habitats considered should be limited. Practical considerations for choosing among methods are suggested.