@incollection{MIP-0707,
author="M. T. Gallegos, G. Ritter",
title="A trimmed version of the EM-algorithm for contaminated mixtures",
institution="Fakult{\"a}t f{\"u}r Informatik und Mathematik, Universit{\"a}t Passau",
year=2007,
number={MIP-0707}
}
We establish a mixture model with spurious outliers and derive its maximumintegratedlikelihood estimator. It may be computed by a trimmed version of the EMalgorithm which we call the EMTalgorithm. We analyze its properties and compute the universal breakdown values of the estimator for normal mixtures. Estimation of the covariance matrix turns out to be robust.