In this paper we investigate the robustness properties of the deepest regre
ssion, a method for linear regression introduced by Rousseeuw and Hubert [6
]. We show that the deepest regression functional is Fisher-consistent fur
the conditional median, and has a breakdown value of 1/3 in all dimensions.
We also derive its influence function, and compare it with sensitivity fun
ctions. (C) 2000 Academic Press.