Consistency in statistical analysis of standardized toxicity studies

Two statistical methods are described and discussed for preparing data from continuous toxicity studies with standard experimental design to be analyz ed statistically by parametric methods. Both methods are intended to bring consistency to the preclinical area of drug development by standardizing th e statistical methods used to assess the results. The first method is a Box -Cox transformation method in which responses are transformed to achieve ho moscedasticity; the second method is a power-of-the-mean modeling of the va riance. The two methods are comparable regarding homoscedasticity and homog eneity across studies. However they have different advantages and drawbacks . The most important advantage of the power-of-the-mean model compared with the Box-Cox transformation is that results from the analyses are on the sa me scale as the original measurements. Hence, the comparability of results from different studies is facilitated. Empirical analyses have been perform ed for different studies and responses to compare and assess the two method s.