Robust Statistics

The traditional setting of statistical inference involves finite or known spaces and subjects. During the past few decades, a theory has been developed that allows the sample space to be abstract or unknown. More recently, mathematical techniques-especially the method of sieves-have been constructed to enable inferences to be made in abstract parameter spaces. This work began with the author's 1950 monograph on inference in stochastic processes (for general sample space) and with the sieve methodology (for general parameter space) This paperback reprint of a Wiley bestseller studies both of these cases and represents the first comprehensive treatment of the subject.