Multi-aspect permutation tests in shape analysis with small sample size

Inferential methods known in the shape analysis literature make use of configurations of landmarks optimally superimposed using a least-squares procedure or analyze matrices of interlandmark distances. For example, in the two independent sample case, a practical method for comparing the mean shapes in the two groups is to use the Procrustes tangent space coordinates, if data are concentrated, calculate the Mahalanobis distance and then the Hotelling T²-test statistic. Under the assumption of isotropy, another simple approach is to work with statistics based on the squared Procrustes distance and then consider the Goodall F-test statistic. Despite their widespread use, on the one hand it is well known that Hotelling’s T²-test may not be very powerful unless there are a large number of observations available, and on the other hand the underlying model required by Goodall’s F-test is very restrictive. For these reasons, an extension of the nonparametric combination (NPC) methodology to shape analysis is proposed. Focussing on the two independent sample case, through a comparative simulation study and an application to the Mediterranean monk seal skulls dataset, the behaviour of some nonparametric permutation tests has been evaluated, showing that the proposed tests are very powerful, for both balanced and unbalanced sample sizes.