R help - May 2008 - Robust Stepwise Regression

Hello, I am interested in performing a stepwise regression using the robust
regression technique to estimate the models at each stage.

At the moment I am using the lm code and the stepAIC code to select the
"best" OLS model. This final model is then re-estimated using the rlm
code.

I am hoping there is a better was to do this, where at each step of the
"stepwise" code the robust regression technique rather than OLS is
used. I realise that an AIC statistic may not be available in rlm, but some
other criteria is fine.

Thanks.


	[[alternative HTML version deleted]]