Parametric robustness: Small biases can be worthwhile (Q797242)

This paper deals with the estimation of the parameters of a Gaussian linear model \(M_ 0\) entertaining the possibility that \(M_ 0\) is invalid and a larger model \(M_ 1\) should be assumed. Estimates are robust if their maximum risk over \(M_ 1\) is finite and the most robust estimate is the least square estimate under \(M_ 1.\) The author applies robustness ideas of Hodges/Lehmann and Efron/Morris to obtain biased estimates which do well under \(M_ 0\) at a small price in robustness. Extensions to confidence intervals, simultaneous estimation of several parameters and large sample approximations applying to nested parametric models are also discussed.