The authors present two nonparametric tests which can be used to compare the level of a random variable based on two independent random samplings. Wilcoxon's (Mann-Whitney's) test is based on the sum of sequences which are appropriate for individual elements in a common group. In large samples the tested criterium is based on the normal distribution, for small samples a table of critical values is enclosed which makes rapid decision possible. The median test is based on the idea that if two independent random samples are from the same population both samples contain equal numbers of elements above and below the median for the common group of all elements of both sub-groups. The tested criterium to which the median test leads is the chi 2-test or Fisher's test of accurate probabilities.

Katedra sociálního lékarství lékarské fakulty Masarykovy univerzity Brno