# Re: st: Interpretation of Curvilinear Effects

 From John Antonakis To statalist@hsphsun2.harvard.edu Subject Re: st: Interpretation of Curvilinear Effects Date Tue, 09 Jun 2009 20:42:43 +0200

Hi:

First, don't use stepwise regression--it is the plague; no worse. Many journals simply won't even review manuscript with such data-driven methods (unless used for a particular goal--ridge, least-angular regression). For instance, see:

Thompson, B. (1995). Stepwise Regression and Stepwise Discriminant-Analysis Need Not Apply Here - a Guidelines Editorial. Educational and Psychological Measurement, 55(4), 525-534.

The best way to get a feel for the shape of the interaction is to plot it; i.e., fit the model, and then plug the numbers across your x-values and plot the predicted value. In the case of a positive x and positive x^2 the line should be relatively flat and positive and the shoot up like a "J" shape.

Or try this after you fit the model:

predictnl y_hat= 1.89 + _b[x]*x + _b[x^2]*x^2 , ci(yhat_left yhat_right)
twoway (connect y_hat  yhat_left yhat_right x, sort)

HTH,
J.

____________________________________________________

Prof. John Antonakis
Associate Dean Faculty of Business and Economics
University of Lausanne
Internef #618
CH-1015 Lausanne-Dorigny
Switzerland

Tel ++41 (0)21 692-3438
Fax ++41 (0)21 692-3305

Faculty page:
http://www.hec.unil.ch/people/jantonakis&cl=en

Personal page:
http://www.hec.unil.ch/jantonakis
____________________________________________________

On 09.06.2009 20:31, Christian Weiss wrote:
Dear Statalisters,

I am currently evaluating curvilinear effects of the variable X on Y
by standard OLS. To asses the curvilinearity I generated X^2 and X^3
and included these variables stepwise in the regressions. Looking at
the significance levels and the signs of the coefficients I concluded
the following:

if X and X^2 is significant:
-> X has a positive sign and X^2 has a negative sign: Bell Shaped
-> X has a negative sign and X^2 has a positive sign: U Shaped

However, the last regression analysis yielded significant X and X^2,
but both with the same sign. I am struggling how to interpret this
effect. Any help is highly appreciated! :)

Best regards
Chris
*
*   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/