Answers are grounded in this paper and cite their sources.
Summary A geometry similar to the “co-ordinate-free” approach to least squares estimation in linear models is developed for R- and M-estimation. This is done by replacing Euclidean distance with appropriate distance measures. This approach allows development of a theory of estimable functions even in cases where expectations of errors do not exist, as well as a simple treatment of general linear hypotheses.
No takes yet. Share an insight, caveat, or question.
McKean et al. (1980) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: