a simple stata program to residualize some variable on some other variables
download this repo.
coming soon:
- installation via
net install.
residualize varname varlist, [replace gen(varname) absorb(varlist)]residualize takes in a varname and residualizes varname on varlist (time series and factor operators are allowed;
this means i.varname syntax is supported).
absorb(varlist) is supported; since we use reghdfe under the hood, interactions of the form absorb(var1 var2#var3) are allowed.
One of replace or gen(varname) is required.
replace overwrites the original variable; gen(varname) saves the residualized variable to varname.
Note: this short program was developed to aid in visualizations. For this reason, after the residualization is performed, the mean is added back to the residuals. In a linear regression, the estimates will be the same with and without the means, since we are just adding or subtracting a constant. However, non-linear specifications are not, in general, robust to such addition of constants.
If you would like to use residualize with a non-linear model, don't.
If you open an issue/feature request for it, I'll implement the nomeans option.
It's on my to-do's but is not particularly high priority (until someone requests it).
coming soon:
- examples + syntax
- implement
nomeans: option to not add means back in