Interpreting a fixed effects regression whether the treatment variable is binary

[Iz Monnick]

Hello all,

There have been similar posts about this interpretation but I can't find any with the same variation of this as I have (but do point me to another topic if I've made a mistake about this).

I am running a fixed-effects linear regression using panel data, which has 9 time periods (9 years). The dependent variable of interest is subjective wellbeing (subwb) and the independent variable of interest is whether the individual lives in an ethnic enclave (blackee_2x). I am confused about how to interpret the coefficient for blackee_2x, considering the temporal aspect of the data. blackee_2x is a binary/dummy variable where 0 is not living in an ethnic enclave and 1 is living in an ethnic enclave.

Ignoring the insignficance of the coefficient and the low F statistic, does the coefficient for blackee_2x = 0.941 (3dp) mean that a change from living outside an ethnic enclave into an ethnic enclave is associated with a 0.941 increase in subjective wellbeing by the end of the panel data set, irrespective of when (within the 9 years) the change occurred?
Or is this interpretation incorrect?

Any help would be much appreciated.

Iz

[Clyde Schechter]

Yes, your interpretation is correct. Obviously the usual caveats about bias in the sample, confounding by unobserved variables that are not time-invariant, measurement error, etc. apply. But that is the basic interpretation of these results.

[Iz Monnick]

Thank you! And yes, those caveats definitely apply.
One more question: does the fixed effect regression use all of the datapoints, including individuals who don't have a change in blackee_2x? Or does it only use those who have a change in blackee_2x?

Edit: just realised this was a silly question and that Stata only uses observations with a change in blackee_2x