Fairlie Interpretation

[Marie Ferrist]

I know that Fairlie computes the difference in Pr( depvar!=0) between the two groups defined by groupvar and quantifies the contribution of group differences in the indepvars to the outcome differential. Furthermore, fairlie estimates the separate contributions of the individual independent variables (or groups of independent variables).

I have two samples. Black and white people. My code is as follows, my independent variable is college education (binary 1 and 0):

fairlie `depvar‘ `indepvars1’ `indepvars2‘ `indepvars4’ `indepvars5‘ `indepvars6’ [pw=wt] if age >= 18 & age <= 65 & employment_2 == 1 & employment_1 == 0 & lagged_health == 0 & black == 1 | white==1 , by(black) pooled(black) ro reps(100)

How should I interpret the Fairlie results after this command? If we consider the statements, it seems that I am not comparing only black and white people to each other? Because in both lag 1 and lag 2 values, we also include the status of having a job and lag health status. Any leads?

[Marie Ferrist]

Any leads would be aprriciated.

[George Ford]

It's Blinder/Oaxaca decomposition for non-linear models. What you read about BO is informative. The Stata Journal articlae on -oacaxa- command usefully describes what's up.

You are trying to account both for race (the race dummy) and the difference in endowments between races. For instance, it may be that there is no difference between races when accounting for other factors.

Say you are looking at salary, and you have male/female and education. While there may be a difference in mean salary between races, it may be entirely explained by differences in education.