A chi-square goodness-of-fit analysis of dependent resource selection data

The chi2 goodness-of-fit test is commonly used for testing if animals use resources in proportion to availability. This method assumes independence of resource selection among animals. In reality, this assumption is violated if animals display antisocial or gregarious behavior. Data from a study of sharp-tailed grouse in eastern Washington suggested some dependency among observations. Realizing that this dependency can have a great influence on inference for resource selection data, we develop a technique to incorporate information on dependent observations through a simple adjustment of the usual goodness-of-fit statistic. We also demonstrate how confidence intervals on proportional use may be modified for dependent observations. Simulation is used to compare our method to other methods.