0
$\begingroup$

I want to use fixed effect model but I want to test an interaction between female(1 if true) which is not changing across users(id). I wrote Stata code shown below and the results are shown after.

IS it correct to do such cross level interaction in a fixed effect model. IF it is significant, how can I interpret the results?

Many thanks if anyone can help.

DV: wage IV: working experience, working weeks, and whether is a female. xtset id t xtreg lwage experience weeks i.female#c.experience, fe

The results is shown below: enter image description here

$\endgroup$
0
$\begingroup$

This is fine. I guess you meant to say that female does not vary across time, $t$, not across users, $id$. Because your experience varies across time, the new variable, $1.female*c.experience$ will not be dropped. The interpretation is standard for an interaction of a continuous and binary variable. The effect of a 1-year increase in experience on wages is 0.48% lower for women compared to men.

$\endgroup$
  • $\begingroup$ Thank you so much for the detailed illustration. It helps a lot! Now I am confident about using such interaction terms. $\endgroup$ – Tim Mar 19 at 21:19

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.