Query about interaction terms in fixed effects panel data

[Abdullah Ijaz]

Hi,

I have a question about fixed effects panel data interaction terms:


Here is my code,

xtreg DepVAr c.IndependentVar#c.InteractionVar controls IndependentVar InteractionVar i.year , fe cluster(code)



My question is related to the coefficient values. So the value of my coefficient for interaction terms in 0.3 significant at P<0.05

The coefficient val of IndependentVar is 0.6 significant at P<0.05 and coeffcient of InteractionVar is insignificant.


Question is:

Is it okay sometimes to report that interaction term is useful even if it is smaller in value as depicted above? I know it can depend on research question econometric technique. But I am just asking whether in the research community, are these results just fine?

P.S. When I exclude interaction term, both my Independent and Interaction Vars become insignificant.

[Carlo Lazzaro]

Abdullah:
without taking a look at what Stata gave you back (I mean the .xtreg, fe- outcome table) I guess that you should leave the interaction in.
It may be that the significant predictor actually leads the statistical significance of the interactions.
Obviously, most depends on other issues, such as model correct specification.

[Abdullah Ijaz]

Thanks a lot fot your feedback. This is how my reg looks like:

[Carlo Lazzaro]

Abdullah:
set aside the recommendations about not posting screenshots but using CODE delimiters (plese, see the FAQ. Thanks), your regerssion code and output look Ok.
It may be that due to limited within-panel variation of time-varying variables, most of your coefficients do not reach statistical significance.
As already wrote, most depends on other issues, such as model correct specification.