Bookmark and Share

Notice: On March 31, it was announced that Statalist is moving from an email list to a forum. The old list will shut down at the end of May, and its replacement, statalist.org is already up and running.


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: st: conflicting results on interaction effect


From   "Casey P. Durand" <durandca@usc.edu>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: conflicting results on interaction effect
Date   Wed, 21 Jul 2010 17:51:20 -0700

Simon,

Yes, the coefficient of X is the relationship between X and Y when and
only when Z=0.  Assuming you want to leave your coefficients in their
raw forms, i.e. not exponentiating them and getting odds ratios, then
to understand what is happening when Z=1, you have to create a new
variable in which you reverse the coding of your original Z variable,
(call it Z_new) such that the old 1 becomes the new 0 and the old 0
becomes the new 1.  You then have to create a new interaction term
with Z_new and X , and then substitute them both into your original
model.  The X coefficient for this new model will now be the effect of
X on Y when Z is equal to 1, using the coding scheme of your original
Z variable.  These two models will then provide the coefficients, p
values and confidence intervals necessary to do a simple slopes
analysis of your interaction.

As for graphing, check out the user written command -predxcon-.  This
provides a very simple way to graph a continuous by categorical
interaction.

Hopefully this is clear enough to make sense.

Casey

On Wed, Jul 21, 2010 at 3:49 PM, Meng Zhao <mrzm_www@yahoo.com> wrote:
>
> Hello,
>
> I run a binomial logistic regression with an interaction variable:
>
>     Coef.         P>z
> x    6.958805      0.026
> z    5.11967       0.000
> x*z -7.490387      0.034
>
> x is a continuous variable, z is a dummy variable. Does this result indicate a sig. positive relationship between x and dependent variable when z is 0, and a negative relationship between x and DV when z is 1?
>
> The plot of this interaction shows that there is a positive relationship between x and DV for both situations (z=1 & z=0). Why this conflicts with the above result?
>
> Thanks.
> Simon
>
>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/statalist/faq
> *   http://www.ats.ucla.edu/stat/stata/

*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   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   |   Site index