Answers are grounded in this paper and cite their sources.
Let Zₙ denote the system of $8n$ independent pairs of measurements (Xᵢₖ, Yᵢₖ), for 4i = 1, 2, ⋯, n and k = 1, 2, ⋯, 8, of two nonobservable random variables ξᵢₖ and ηᵢₖ, known to satisfy a linear relation of the form ξᵢₖ cos θ^ + ηᵢₖ sin θ^ - p = 0, where p is an arbitrary real number and θ^ may have any value between the limits -1/2π < θ^ 1/2π. The purpose of the paper is to construct a class of estimates Tₙ(Zₙ) of the parameter θ defined as follows: when θ^ = 1/2π then θ = 0; otherwise θ ≡ θ^. Each estimate Tₙ(Zₙ) of the class considered converges in probability to θ as n → ∞ under the following conditions: (i) except when θ = 0, the variables ξᵢₖ are nonnormal; (ii) any nonnormal components of the errors of measurements, Xᵢₖ - ξᵢₖ and Yᵢₖ - ηᵢₖ, are mutually independent, independent of ξᵢₖ and of the normal components of these errors; (iii) the normal components of the errors may be correlated but as a pair are independent of ξᵢₖ.
No takes yet. Share an insight, caveat, or question.
Jerzy Neyman (1951) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: