chb(eff=NULL, ch=<<see below>>, cb=<<see below>>)
1) giving the desired asymptotic efficiency of a
location M-estimate at the Gaussian model.
Only values between
0.7 and
1.0 are allowed.
If
eff is supplied, then the input values of
ch and
cb are ignored.
ch is supplied, then
cb must also be supplied.
cb is supplied, then
ch must also be supplied.
34 by
5 matrix is returned.
The columns are: efficiency (1), Huber tuning constant (2),
Huber beta value (3), bisquare tuning constant (4), and bisquare beta (5).
ch.
A beta parameter is the expectation of the square of a psi function under
the Gaussian distribution, and is used to achieve statistical consistency
at the Gaussian model for the scale estimate in
robloc.
Hampel, F. R., Ronchetti, E. M., Rousseeuw, P. J. and Stahel, W. A. (1986).
Robust Statistics: The Approach Based on Influence Functions.
Wiley, New York.
Huber, P. J. (1981).
Robust Statistics.
Wiley, New York.
chb(.95)$ch # tuning constant for the 95% efficient Huber M-estimate of location. chb(ch=1.345, cb=4.685)