A nonparametric approach to the statistical analysis of mutagenicity data.

A nonparametric statistical method for analysing count data from mutagenicity tests is proposed. This method does not make any stringent assumptions on the variation of the data. It requires only one ranking of the observations and the calculation of a resulting test statistic. The significance of this statistic can be assessed by approximation to the standard normal distribution or tables of critical values calculated by the authors. The method also includes a descriptive component through a series of K-1 estimates of a trend in a K group experiment. Several examples are discussed to illustrate the proposed method.