ABSTRACT
Estimating a location parameter of multivariate observations is a very common and important task in the practice. If we want to estimate the location robustly, we try to apply similar robustness criteria and measures of performance of estimators, such as in the univariate model. We characterize the estimators by the influence function, global and local sensitivities, breakdown point, the maxbias and a high asymptotic efficiency under the normal distribution. We also desire to have an estimator equivariant in a suitable sense.
