Statistical Methods for Comparative Analyses

Suny At Stony Brook, Stony Brook NY

Investigators

Abstract

0212023 Rohlf Ordinary statistical methods cannot be used to study variation or correlations among species because the fundamental assumption of independence of observations is violated due to the pattern of similarity expected as a result of their common evolutionary history. This project is concerned with the development and evaluation of special statistical methods that can be used to take this dependence into account. Work is proposed on generalized least-squares models, mixed effects models, generalized linear models (using extensions for non-independent data), and generalizations for the analysis of multivariate data (especially for suites of shape variables obtained from geometric morphometric studies). Generalizations to models of reticulated evolution and geographic sampling will also be investigated. These methods will be applied to several data sets in order to evaluate their usefulness in practice. Support for a graduate student is included in the project, along with enhancements to software programs that the Principal Investigator has made available for many years to the science community.