Generate a Robust Family Object
DESCRIPTION:
Generates a new family object, with its component functions suitably modified to perform a robust version of the glm or gam fit.
USAGE: 
robust(family = gaussian, scale=0, k=1.345, maxit=10)
OPTIONAL ARGUMENTS:
family:
a family object, as produced by gaussian, binomial, poisson, etc.
scale:
a scale parameter can be supplied. If omitted, a robust scale is estimated based on a scale of 1 and the fit of the first glm or gam iteration.
k:
a threshold used in the computation of the robustness weights.
maxit:
the maximum number of iterations.
VALUE:
a new family object, with its component functions suitably modified to perform a robust version of the glm or gam fit.
DETAILS:
Provides a method not greatly affected by outliers. Gives a good fit to the bulk of the data.
SEE ALSO:
glm , gam , family , power , family.object .
EXAMPLES: 
# Fit a linear model robustly
glm(formula, family=robust)
# Fit a generalized additive model robustly
gam(formula, family=robust(binomial))