METHODS OF ASSESSING AND ACHIEVING NORMALITY APPLIED TO ENVIRONMENTALDATA

It has been recognized for a long time that data transformation method s capable of achieving normality of distributions could have a crucial role in statistical analysis, especially towards an efficient applica tion of techniques such as analysis of variance and multiple regressio n analysis. Normality is a basic assumption in many of the statistical methods used in the environmental sciences and is very often neglecte d, in this paper several techniques to test normality of distributions are proposed and analyzed. Confidence intervals and nonparametric tes ts are used and discussed. Basic and Box-Cox transformations are the s uggested methods to achieve normal variables. Finally, we develop an a pplication related to environmental data with atmospheric parameters a nd SO2 and particle concentrations. Results show that the analyzed tra nsformations work well and are very useful to achieve normal distribut ions.