Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: Selecting Best Regression Equation using STATA


From   Ronán Conroy <rconroy@rcsi.ie>
To   "statalist hsphsun2.harvard.edu" <statalist@hsphsun2.harvard.edu>
Subject   Re: st: Selecting Best Regression Equation using STATA
Date   Tue, 06 Apr 2004 11:19:27 +0100

on 05/04/2004 19:34, wildscop at appstat@hotpop.com wrote:

> 
> Does STATA provide any command to form any of the following procedures to
> find Best Regression Equation -

Stata's strength lies in its user-contributed procedures. In this case, the
best regression equation is of the form

Response variable is a function of the theory you are using informed by
interacting with the data, adjusted for confounding variables.

I don't mean to be flippant, but the construction of a regression equation
is like the investigation of a crime. Many leads need to be checked out, and
it has to fit the known data but, most important of all, it has to make
sense. The trouble with 'automatic' methods is that they don't know any
science, and cannot be expected to.

And what it there is no science? What if you just have a whole bunch of
predictors and want to weed out the redundant ones? The trouble is model
shrinkage. The model you build will over-estimate the ability of the
predictor variables. If applied to a new sample of data, the fit of any
model will drop by a factor that is the ratio of adjusted-R-squared to
R-squared (for least squares models - other models the calculation is more
complex). However, for stepwise models and best subset models the adjusted
R-squared should be calculated based on *all* the predictors that were
*considered* for inclusion, not just the ones that were included. With many
predictors, this can mean that your model shrinkage makes it useless a
priori. 



Ronan M Conroy (rconroy@rcsi.ie)
Lecturer in Biostatistics
Royal College of Surgeons
Dublin 2, Ireland
+353 1 402 2431 (fax 2764)

--------------------
Just say no to drug reps
http://www.nofreelunch.org/

--------------------------------------------------------------------------------------------------------------------
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom
they are addressed.
If you have received this email in error please notify the
originator of the message. This footer also confirms that this
email message has been scanned for the presence of computer viruses.

Any views expressed in this message are those of the individual
sender, except where the sender specifies and with authority,
states them to be the views of The Royal College Of Surgeons in Ireland.

--------------------------------------------------------------------------------------------------------------------
*
*   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