A method for normalizing drug responses to enhance reliability of parametric statistical tests
- Additional Document Info
- View All
Experimental constraints often require that pharmacologists study the effects of treatments upon responses elicited with a single concentration of agonist. The choice of statistical analysis, nonparametric or parametric, will depend upon whether or not the responses are normally distributed. This study uses both Monte Carlo and theoretical approaches to examine the normality assumption as it applies to drug responses. Responses measured in systems where drug sensitivities follow a lognormal distribution are generally not normally distributed. A simple transformation of responses, however, can restore normality. Researchers may find it beneficial to transform response data prior to performing t tests, analysis of variance or other parametric statistics in order to preserve the power and confidence level of the test.
author list (cited authors)