The multi-purpose permutation procedures (MRPP) test statistic is desi
gned to analyze multivariate data at the ordinal or higher levels. It
is based on the weighted averages of symmetric distance functions over
all paired objects within a priori disjoint groups of objects from a
finite population of objects where each object's response is a point i
n an r-dimensional space. Thus, the arguments of each r-dimensional po
int correspond to the r measured responses of each object in the finit
e population of objects. The null hypothesis underlying MRPP is that t
he observed sample of objects within groups of a specified size struct
ure is randomly obtained from the pooled collection of objects compris
ing the finite population. The procedure is used to test the presumed
geomorphic differences among three reaches of the Lower Mississippi Ri
ver. A combination of the proposed reaches is favored and several vari
able-based reach configurations are proposed.