Classical Analysis of Variance Methods and Nonparametric Counterparts

The present investigation is concerned with the use of classical and nonparametric techniques in the analysis of experiments. Occasionally biometricians are confronted with results arising from an experiment with data which do not necessarily satisfy the assumption of Normality, the basis of the classical analysis of variance methods. Sometimes the biometrician handles such a situation by transformation of the observations and applying the classical techniques. Although this attack may lead to a satisfactory analysis, the results may be questionable in a number of cases. In analyzing continuous data without the Normality assumption nonparametric methods provide realistic alternatives. In this paper a number of problems will be discussed with and without the Normality assumption resulting in the well‐known classical analysis and its nonparametric counterparts. Most of the nonparametric procedures to be discussed are based on ranks.