A simple robust regression method is the M-estimator, which
systematically reduces the effect of data on the basis of the sizes of
residual elements. M-estimators may not be sensitive to leverage
points, depending on whether they produce unusual residuals
(Chave and Thomson, 2003). In fact, as M-estimators do not account
for leverage points, they have unbounded influence. In other
words, they do not bound outliers in the x-direction (Hampel et al.,
1986). To overcome this problem, the use of bounded influence
estimators is suggested. The underlying purpose of bounded in-
fluence estimators is to form weights that account for outliers in
both the y- and x-directions