I was reading about a step-down procedure in Harrell's Regression
Strategies book. It was a method for taking a full regression model and
identifying reduced set approximations of the model that would account for
specified amounts of variance as the full model (.eg., 95%). Essentially,
the fitted values of the initial model are regressed upon the predictors and
then the model is paired down from there based upon the relative
contribution of predictors. He suggests that this is one potential
alternative to developing more parsimonious models that avoids many, but not
all, of the problems with stepwise models. He shows how to implement this in
S-plus. I was wondering if there is a way to do it in Stata.
*
* For searches and help try:
* http://www.stata.com/support/faqs/res/findit.html
* http://www.stata.com/support/statalist/faq
* http://www.ats.ucla.edu/stat/stata/
