Re: st: Interaction terms when one variable is omitted - query

[Maarten Buis <maartenlbuis@gmail.com>]

If you want to know by how much the effect of job_loss differes
between those with and without a college degree then you can just
interpret your interaction terms as you would in a normal model. In
your model you (secretly) included the main effect of college as part
of your fixed effects.

Hope this help,
Maarten


On Thu, Apr 18, 2013 at 6:35 PM, Mirnezami, Oliver
<O.Y.Mirnezami@warwick.ac.uk> wrote:
> Hello
>
> I have a simple model with an interaction term but one of the covariates is omitted and so I'm not sure how to then interpret the interaction term.
>
> e.g.
>
> health = constant + job_loss_dummy + college_degree_dummy +  job_loss_dummy x college_degree_dummy  + other_controls + error term
>
> As this a fixed effects model, the college degree dummy is omitted as it is time-invariant (this uses retirement study data so everyone got their degrees many years ago so it does not change over time within the sample).
>
> The interaction term is significant and the job loss dummy is significant as well.
>
> I would be grateful for any advice on how to interpret the interaction term.
>
> I have looked at various help pages online and textbooks but haven't found an example of this nature where one of the covariates is omitted (in this case due to fixed effects).
>
> Thank you
>
> Oliver
>
>
> *
> *   For searches and help try:
> *   http://www.stata.com/help.cgi?search
> *   http://www.stata.com/support/faqs/resources/statalist-faq/
> *   http://www.ats.ucla.edu/stat/stata/



-- 
---------------------------------
Maarten L. Buis
WZB
Reichpietschufer 50
10785 Berlin
Germany

http://www.maartenbuis.nl
---------------------------------
*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/