Bookmark and Share

Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at

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

Re: st: Quadratic Function, Turning Points and Interaktion Terms

From   Maarten Buis <>
Subject   Re: st: Quadratic Function, Turning Points and Interaktion Terms
Date   Wed, 9 May 2012 10:19:41 +0200

On Tue, May 8, 2012 at 9:54 PM, Lukas Borkowski wrote:
> I work on a model that can be simplified as y = a_x1 b_x2 c_x3
> My first findings suggest that the effects of x1-3 on y will change with increases in x1-3. Now, do I have to run
> reg y x1 x1sq
> di _b[x1]/(2*_b[x1sq])*-1
> for every x seperately or combined, thus
> reg y x1 x1sq x2 x2sq x3 x3sq
> in order to find the correct turning points?

That depends on whether or not you want to the turning points adjusted
for the other variables. There can be no general answer to that, it
just depends on your problem, your question, your theory, etc.

> What happens when I add interaction terms, lets say for g1 and g2 for groups.
> Should I run the regression over my unrestricted equation including all interaction terms?

That depends on what you want to do and how much information is
present in your data.

Rather than trying to find "The Answer" I would look substantively at
what I want to learn from the data, what information is present in the
data, and how I can combine the two in a sensible argument.

You may also find the following link useful:

Hope this helps,

Maarten L. Buis
Institut fuer Soziologie
Universitaet Tuebingen
Wilhelmstrasse 36
72074 Tuebingen
*   For searches and help try:

© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   Site index