Abstract

This presentation suggests new semi-parametric correlations to measure the dependency in a bivariate circular-circular or a circular-linear sample. These measures are related to Mardia’s circular rank correlation. Their large sample distributions are derived using theorems on the asymptotic behaviour of multivariate rank statistics. Two applications are presented. First a new semi parametric goodness of fit test for the Wehrly and Johnson circula is proposed. Then good of fit tests and semi-parametric estimators for the parameters of a copula model for circular-linear association are proposed. Data sets on animal movement in ecology are analyzed using the proposed methodology.