A simple nonparametric procedure: the MAXIMIN test.

The MAXIMIN test is based on the principle that if two samples were drawn from the same population much overlapping of the values of the two groups would occur. However, if the two samples are drawn from two different populations, the values of Group 1 are expected to accumulate at one end of the numerical scale, whereas those of Group 2 will accumulate the other end. The sum of nG1 (the number of values in Group 1 greater than the greatest value in Group 2) and nG2 (the number of values in Group 2 smaller than the smallest value in Group 1) gives the MAXIMIN value (MV). The higher MV is, the smaller is the probability that the two samples contain different values by chance. The MAXIMIN test may provide a simple alternative to the t test as well as to other parametric and some nonparametric procedures for evaluating clinicopharmacological results.