Statalist The Stata Listserver


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

st: RE: standardized parameter for dichotmious variable in a regr


From   Steven Samuels <ssamuels@albany.edu>
To   statalist@hsphsun2.harvard.edu
Subject   st: RE: standardized parameter for dichotmious variable in a regr
Date   Wed, 23 Aug 2006 13:26:06 -0400

I agree with Maarten.

The beta coefficients measure impact on the conditional MEAN of Y. If you believe that r-square measures strength of
association in the simple X-Y case, then you should use the partial r square in multiple regression.

This works for dichotomous and continuous covariates. The partial correlations are ranked in
the same order as the absolute value of the t-statistics for each covariate.

Think of a simple X-Y regression. You can have a "strong" X-Y association (as indicated by the SAME high R-square)
in two ways, holding the X's constant.

1. Large slope but relatively high residual SD.
2. Small slope but small residual SD.

In fact, holding the X's constant, one regression plot can be transformed into the other by changing the
units of Y. Only r-square is invariant to this kind of transformation in the Y's (or in the X's, for that matter.)


Steve Samuels
*
* 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