Statalist The Stata Listserver


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Re: stepwise like SPSS?


From   Ronán Conroy <rconroy@rcsi.ie>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Re: stepwise like SPSS?
Date   Thu, 15 Feb 2007 10:27:56 +0000

On 14 Feabh 2007, at 23:13, David Zatz wrote (concerning doubts about the usefulness of stepwise methods):


Yes, I've seen that about a hundred times on my Google search. However, for
exploratory research, used with care, it is desirable. Sometimes you want
to be guided by what's there and not by what's already in your head.
When you have more candidate predictor variables than you know what to do with, another route is to examine these for a simpler underlying structure. The typical case in which the candidate predictors are widely correlated, which causes problems in OLS models and especially in stepwise ones, can be turned to an advantage in this way.

I've spent the last couple of years trying to figure out some way of cracking a problem with 36 candidate predictors whose factor structure didn't look like it was supposed to (and didn't look like anything I could recognise from the theoretical point of view). I've finally started making sense of the mess using -clv- to explore the structure of the predictors and -msp- to winnow out the less useful ones, leaving me with something which is unexpected but fascinating…


=========
Ronán Conroy
Royal College of Surgeons in Ireland
rconroy@rcsi.ie
+353 (0) 1 402 2431
+353 (0) 87 799 97 95
http://www.flickr.com/photos/ronanconroy




*
* 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/




© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index