[R] testing the signifiance of variables in a robust regression with M-estimators

From: Nicolas Ferrari <nicolas.ferrari_at_m4x.org>
Date: Tue 31 Jan 2006 - 09:42:04 EST


I have computed robust regressions with M-estimator using rlm(). The score function used is psi(x)=c*x/(x^2+c). Distributions are highly concentrated and have very thick tails.

I have not introduced fixed effects because the important number of observations could make numerical difficulties. Nevertheless, I want to test that there is no fixed effect, i.e. I want to test that if I add fixed-effect, their coefficients are not significant. With the form of distributions of residuals (far from normality), I can't used classic tests like Fischer tests.

Do you know what sort of tests I can conduct and how to implement them in R?

Thank you for your reply?

Nicolas Ferrari

        [[alternative HTML version deleted]]

R-help@stat.math.ethz.ch mailing list
https://stat.ethz.ch/mailman/listinfo/r-help PLEASE do read the posting guide! http://www.R-project.org/posting-guide.html Received on Tue Jan 31 09:52:48 2006

This archive was generated by hypermail 2.1.8 : Tue 31 Jan 2006 - 15:01:23 EST