摘要：The Box-Cox family of transformation is a well-known approach to make data behave accordingly to assumption of linear regression and ANOVA. The regression coefficients, as well as the parameter defining the transformation are generally estimated by maximum likelihood, assuming homoscedastic normal error. In application of ANOVA for hypothesis testing in biostatistics science experiments, the assumption of homogeneity of errors often is violating because of scale effects and the nature of the measurements. We demonstrate a method of transformation data so that the assumptions of ANOVA are met (or violated to a lesser degree) and apply it in analysis of data from biostatistics experiments. We will illustrate the use of the Box-Cox method by using MINITAB software.
