Test whether coefficient is significantly different in 2 subsamples

[Peter Meier]

Dear listers,

I want to test whether the relationship of interest is significantly different between the subsamples with respectively var1 above or below its median, and likewise for var2 and var3.
Hence I generated the respective dummies and wrote the below code:

eststo clear
foreach var of varlist dummy1 dummy2  dummy3 {
     eststo: reghdfe depvar indepvar if `var'==0
     eststo: reghdfe depvar indepvar if `var'==1     
     }
esttab using "table"

This does produce the estimates of interest, showing for each dummy first the sub- and then right to it the super-median estimations.
However, I would like to add to the bottom of the table the variable by which I have split and, more importantly,
add a formal test of whether the difference in the estimates on indepvar is significant.

Does someone know how to do this?

Thank you so much,
PM

[Felix Bittmann]

To add notes, esttab provides the option

addnotes("Something to print")

Regarding the other question, you need to add an interaction term. To fully replicate what you have so far would boil down to this model:

reghdfe depvar c.indepvar##i.dummy

The significance level of the interaction term is what you are interested in.

If you have more control variables, you can do it like so:

reghdfe depvar i.dummy##(c.indepvar c.control1 c.control2)