Answers are grounded in this paper and cite their sources.
(i) a distinction must be made between the linear regression equation of a variable y on a second variable x, and a linear functional relation between two variables Y and X masked by errors. The former equation is still available for prediction even if the variable x is subject to error, but is not necessarily appropriate for a functional relation when one exists. (ii) it is possible to set up maximum likelihood equations for the second problem, but they do not lead to a unique solution without further assumptions, such as an assumption about the relative magnitude of the errors in x and y.
No takes yet. Share an insight, caveat, or question.
M. S. Bartlett (1949) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: