Answers are grounded in this paper and cite their sources.
Let Tₙ be an M-estimator with defining function ψ and preliminary estimate of scale sₙ. Without loss of generality, let sₙ → 1 and take Eψ(X/ξ) = 0. Under various conditions, it is shown that any consistent version of Tₙ is almost surely to order O(n⁻¹ log₂ n) a linear combination of n⁻¹ ∑ⁿ₁ ψ(Xᵢ) and sₙ. Only in the case EX₁ψ'(X₁) = 0 does the contribution of Sₙ vanish; it is shown how this affects the estimation of the asymptotic variance of Tₙ.
No takes yet. Share an insight, caveat, or question.
Raymond J. Carroll (1978) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: