Stata The Stata listserver
[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: interactions

From   Gene Fisher <>
Subject   Re: st: interactions
Date   02 Jun 2003 15:29:35 -0400

It seems to me that the effect of normal hours is contigent on
perception of fairness.  That is why the interaction coefficient is
significant.  The significant coefficient of normal hours indicates that
it has an effect when fairness conditions are not present.  perception
of fairness may reduce that effect to zero (I'll bet the interaction
coefficient is about the same size as the normal hours coefficient, but
opposite in sign), so that the coefficient of normal hours is not
significant when an interaction term is not included in the equation. 
Run two regressions, one when fairness is 0 and one when fairness is 1. 
I'll be the coefficient of normal hours flips (and in that case neither
may be significant) or one is strong and the other is near zero (so that
the average of the two is too small to be significant when the data are
pooled).  I don't think the significance of all three coefficients is
suspicious.  Do the coefficients, particularly their signs, make sense?

On Mon, 2003-06-02 at 13:27, Maureen Paul wrote:
> Hi 
> I wonder if anyone can help me figure this out. I am running a fixed effects regression. The dependent variable is overtime hours. On the left hand side I have normal hours of work and a dummy variable indicating fairness perceptions. Both of those variables entered separately appear as insignificant. However, when I interact them, they are highly significant (both the interactions and the separate variables). Isn't there something suspicious about this?
> Thanks for any help on this. I have been thinking about this all day but can't figure it out.
> M
> *
> *   For searches and help try:
> *
> *
> *
Gene A. Fisher
Department of Sociology
University of Massachusetts
200 Hicks Way
Amherst, MA  01003
Phone: (413) 545-4056
Fax:   (413) 545-3204

*   For searches and help try:

© Copyright 1996–2019 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index