Fournier, N.2011-12-162011-12-162011-12-16201110.1016/j.jspi.2011.02.018https://infoscience.epfl.ch/handle/20.500.14299/74052WOS:000291067400014In this paper, we investigate the construction of compromise estimators of location and scale, by averaging over several models selected among a specified large set of possible models. The weight given to each distribution is based on the profile likelihood, which leads to a notion of distance between distributions as we study the asymptotic behaviour of such estimators. The selection of the models is made in a minimax way, in order to choose distributions that are close to any possible distribution. We also present simulation results of such compromise estimators based on contaminated Gaussian and Student's t distributions. (C) 2011 Elsevier B.V. All rights reserved.Pitman estimatorsModel averagingDistance between distributionsLocationLikelihoodtext::journal::journal article::research article