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 on April 23, and its replacement, statalist.org is already up and running.


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

Re: st: Fwd: How to interpret regression?


From   "JVerkuilen (Gmail)" <jvverkuilen@gmail.com>
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: Fwd: How to interpret regression?
Date   Tue, 25 Sep 2012 08:31:09 -0400

On Tue, Sep 25, 2012 at 8:05 AM, Kresten Buch <krestensb@gmail.com> wrote:
>
> Dear Statalist,

<snippety doo dah>

> If I include all variables in the model, I get significant effects
> from both sets of interaction variables. If only one set is included,
> then I nothing is significant, no matter witch one is excluded.
>
> I hope the explanation makes sense ?


This is a very complicated model. All the interactions are likely to
create collinearity issues and make individual regression coefficients
unlikely to be statistically significant. How much does the R^2 change
when you add variables as a block? If it's a lot you can use -test- to
determine if the addition of the interactions is jointly significant,
even if it's hard to tell which one is from the individual
coefficients.

I think that, at least for interpretational issues you should create
some simpler versions as benchmarks to evaluate the interactions
against.

I would also recommend centering all continuous predictor variables
and creating interactions from them. That's statistically equivalent
but should help break up the collinearity effect on the individual
regression weights.
*
*   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