Storing estimates from bysort: reg command into a new variable

[Thomas Jesperson]

Hello,

First time Statalist user here (please let me know if there is any way to improve my post)!
I am using individual-level panel data that includes variables for income, years of education, and occupation-industry. I use the below code to estimate the returns of education on income, by industry-occupation.

bysort occupation_industry: reg income education, r

I would like to store all of the coefficients for education in a variable, but I was not able to find any options in the regress command that allow me to do this. I would usually use _b[varname] or estimates save to save coefficients but as far as I am aware, these are not compatible with bysort: regress.

Any suggestions?

[Nick Cox]

. sysuse auto, clear
(1978 automobile data)

. statsby, by(foreign) : regress mpg weight, r
(running regress on estimation sample)

      Command: regress mpg weight, r
           By: foreign

Statsby groups
----+--- 1 ---+--- 2 ---+--- 3 ---+--- 4 ---+--- 5 
..

. l

     +---------------------------------+
     |  foreign   _b_weight    _b_cons |
     |---------------------------------|
  1. | Domestic   -.0059751   39.64697 |
  2. |  Foreign    -.010426    48.9183 |
     +---------------------------------+

There are options to save more results.

[Thomas Jesperson]

statsby seems to do exactly what I wanted, so thank you. Unfortunately, I failed to mention in my original post that my equation uses pweights, which is the only type of weight that is not supported by statsby. A more correct version of my code(before using statsby) is the following.

  
 bysort occupation_industry: reg income education [pw=sw], r

Do you have any suggestions for a way to include a sample weight when pweights is not allowed? Thanks!

[Nick Cox]

Sorry, no.