On bad data suppression in estimation

Merrill and Schweppe [1] propose a modification of the usual least squares criterion for estimation in order to suppress bad data. It is shown that similar estimators can be obtained from maximum likelihood theory.