Least squares regression vs. geometric mean regression for ecotoxicology studies

摘要

Least squares linear regression and geometric mean regression analysis procedures are compared for use in analyzing ecotoxicology data. These data are characterized by both response data and environmental factor data being measured with error. According to cited literature, when factor data are measured with error, regression analyses for response as a function of environmental factors should be performed using geometric mean regression. A case is made in the current paper that it is more appropriate to use least squares regression in analyses of such data rather than geometric mean regression.