Notice: On April 23, 2014, Statalist moved from an email list to a forum, based at statalist.org.
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: st: radical change in t-stat, sign and significance
From
Nick Cox <[email protected]>
To
[email protected]
Subject
Re: st: radical change in t-stat, sign and significance
Date
Fri, 1 Apr 2011 19:18:30 +0100
This need not be worrying. The signs are only part of the information.
The magnitudes are important too. Also, remember that x and x^2 are
correlated and may to some extent serve as proxies for each other,
even if there is no declared problem of multicollinearity.
What you can easily do is plot the two terms using -twoway function-.
Your intuition can easily mislead. Also, standard calculus methods can
show you the position of the turning point, which may not be within
the range of the data.
Made-up examples can be instructive too, e.g.
twoway function -x, ra(0 10) || function -0.25 * (x^2) + 0.8 * x , ra(0 10)
Nick
On Fri, Apr 1, 2011 at 6:59 PM, Fabio Zona <[email protected]> wrote:
> Dear all,
>
> I have a regression (zero inflated negative binomial): when I include the linear predictor alone (without its square term), the coefficient of this linear predictor is negative and significant.
> However, when I introduce the square term of the same predictor: a) the linear one changes its sign, becomes positive, and it is still significant; b) the square term gets a negative sign and is signficant.
>
> Is this radical change in sign and significance of the linear coefficient a signal of some problems in the model?
>
> the collin command says I have no prob lem of multicollinearity
*
* 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/