Statalist The Stata Listserver


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

Re: st: Obtaining beta coefficient for an interaction term


From   "Austin Nichols" <austinnichols@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Obtaining beta coefficient for an interaction term
Date   Thu, 1 Mar 2007 11:49:10 -0500

Emma--
I think this question is ill-formed or incomplete.  There are various
methods of "standardizing" coefficients, and you can
view http://www.indiana.edu/~jslsoc/stata/spost9_ado/listcoef.hlp
to see several in Stata.

But if you want to "standardize" a linear combination of coefficients
(not an interaction term, to be sure, but also not bX + CX^2, I am
guessing), you will need to add some specificity.

Let us suppose you want to "standardize" the marginal effect of X, given by
b + 2cX
(the partial of your bX + CX^2) in some model.  Do you mean to divide
that combination by the estimated standard deviation of X to interpret
the marginal effect in standard deviation units? At what value of X?
Hard to see exactly what you actually want here, but it may be
-lincom-.

sysuse auto, clear
g mpg2=mpg^2
reg pr mpg*
su mpg if e(sample)
local m=r(mean)
local s=r(sd)
lincom (_b[mpg]+2*_b[mpg2]*`m')/`s'

On 3/1/07, emmmasa <emmmasa@yahoo.com> wrote:
Dear all:
Could anyone advise how to derive the overall beta
(standardized) coefficient for an interaction term
(e.g., bX + CX^2) for a standard OLS regression?
Thanks very much.
Emma
*
*   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