* open data
use cps4_small, clear
describe
summarize

* estimate model with black-female interaction
reg wage educ i.black##i.female

* estimate wage difference between black-female and white-male
lincom 1.black + 1.female + 1.black#1.female

* F-test of joint significance
test 1.female 1.black 1.black#1.female

* Average marginal effects
margins, dydx(*)

quietly summarize black
scalar ablack = r(mean)
lincom 1.female + 1.black#1.female*ablack

* Chapter 7.2.2 Add regional indicators
reg wage educ i.black##i.female i.south i.midwest i.west
test 1.south 1.midwest 1.west

di "F(3,992,.95) = " invFtail(3,992,.05) 
di "F(3,992,.90) = " invFtail(3,992,.10)

* Chapter 7.2.3 Testing the equivalence of two regressions
reg wage i.south##(c.educ i.black##i.female)
test 1.south 1.south#c.educ 1.south#1.black 1.south#1.female ///
     1.south#1.black#1.female

* constructing estimates in separate regressions from fully interacted model
lincom 1.black + 1.black#1.south
lincom 1.female + 1.female#1.south

* Estimate separate regressions
bysort south: reg wage educ i.black##i.female