Re: st: Standardized interaction terms - which p-values hold?

[Maarten Buis <maartenlbuis@gmail.com>]

On Tue, Jan 15, 2013 at 5:59 PM, Elisabeth Bublitz wrote:
> This leaves one more question open: How should you handle changing p-values?
> The so far prefered option (A) returns something different than a regression
> with unstandardized variables (B).
>
> From what I know, a standardization of variables should only change
> coefficients not p-values.

That is not true when your model includes interactions. Say your model
has two variables: age (A) and breath (B) and an interactionterm A*B.
The effect of age is the effect of age when breath = 0 and similarly
the effect of breath is the effect of breath when age = 0. So if you
change the meaning of the value 0 of your variables by standardizing
the variables age and breath, than that involves a substantive change
in the null-hypothesis and should thus result in a change in p-values.

So the solution is, as always, to think carefully about which null
hypothesis you want to test before reporting p-values.

> To get identical p-values as in the baseline
> regression (with unstandardized variables) it is necessary to leave
> variables unstandardized before creating the interaction. If I then
> standardize the interaction term, it gives identical p-values (C).

As I explained above, that is true, but typically that is not what you want.

-- Maarten

---------------------------------
Maarten L. Buis
WZB
Reichpietschufer 50
10785 Berlin
Germany

http://www.maartenbuis.nl
---------------------------------
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/