Abstract
The consistency of the least trimmed squares estimator (see Rousseeuw [14] or Hampel et al. [8]) is proved under general conditions. The assumptions employed in paper are discussed in details to clarify the consequences for the applications.