On the breakdown point of multivariate location estimators based on trimming procedures (Q808584)

For constructing location estimators in multivariate cases based on trimming procedures, the author defined impartial \(\alpha\)-trimmed \(\Phi\)-means and impartial \(\alpha\)-trimmed Chebyshev centres. Results concerning the finite sample breakdown points of estimators are also given. It is shown that the breakdown point of impartial \(\alpha\)-trimmed \(\Phi\)-means and Chebyshev centres is always at least loo \(\alpha\) \%, and that for \(\Phi\)-functions satisfying certain conditions the breakdown point reaches maximal 50 \%.