Re: st: interpretation of squared term

[Sarah Elizabeth Edgington <sedging@ucla.edu>]

.
The basic intuition here is that you're generating a quadratic 
curve.  If you have a positive effect of age and a negative effect of 
age squared that means that as people get older the effect of age is 
lessoned.  A positive effect of age and a positive effect of age 
squared means that as people get older the effect is stronger.  You 
probably want to graph it once to get a sense of the pattern even if 
you don't present those graphs.
Meanwhile have you considered breaking age into categories (or using 
individual age dummies if your sample size is sufficient)?  A 
quadratic is less restrictive that a simple linear specification, but 
it is still a restriction.  You may not be accurately capturing the 
actual age pattern with your model.  Of course a linear spline is 
also a possible choice.  Depending on your discipline, though, that 
may not be the conventional solution.

-Sarah


At 12:39 PM 8/26/2010, David Bai wrote:
> Thank you, Maarten.
>
> I include the squared term because this year variable might be 
> non-linearly related to the outcome. Is there an easier way to 
> roughly interpret the year effect? I do not want to go deeper to 
> have an exact interpretation of the two year variables. E.g., what 
> does a negative coefficient for a squared term mean? Thank you.

*
*   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/