Storing multiple coefficients from a regression with interaction term

[Melanie Schoepfer]

Hi all!


I am trying to store the coefficients of the following output under different names:

reg  c.ptile c.ptileFather##t

      Source |       SS           df       MS      Number of obs   =       857
-------------+----------------------------------   F(3, 853)       =      3.72
       Model |  9227.91757         3  3075.97252   Prob > F        =    0.0112
    Residual |  704689.473       853  826.130684   R-squared       =    0.0129
-------------+----------------------------------   Adj R-squared   =    0.0095
       Total |  713917.391       856  834.015644   Root MSE        =    28.742

---------------------------------------------------------------------------------
          ptile |      Coef.   Std. Err.      t    P>|t|     [95% Conf. Interval]
----------------+----------------------------------------------------------------
    ptileFather |  -.0203642    .053272    -0.38   0.702    -.1249238    .0841953
            1.t |   .8206032   4.124634     0.20   0.842    -7.275018    8.916224
                |
t#c.ptileFather |
             1  |  -.0997823   .0693683    -1.44   0.151     -.235935    .0363703
                |
          _cons |    54.0756   3.289373    16.44   0.000     47.61939    60.53182
---------------------------------------------------------------------------------

ideally I'd like to store the coefficients of

• ptileFather as a_1
• 1.t as a_2
• t#c.ptileFather as a_3
• _con as a_0

I've tried to run

gen a_2=.
qui reg c.ptile c.ptileFather##t
replace a_2=_b[x2]

to start off, but Stata does not find [x2]. Is there another way...?

Thanks a lot for your time!

Best,
Melanie

[Rich Goldstein]

all the coefficients are saved by Stata in several places but, at least in your regression, none of them are called "x2" so of course Stata cannot find it; however, you can find what is there by looking at "r(table)" which is automatically saved by Stata after each regression; to be sure you don't lose it, you should probably make a new matrix that is just a copy of it and then extract what you want - why do you want a coefficient to be a variable as that also makes no (little) sense to me? The existence and use of r(table) has been discussed in this forum many times and a search on that term should help

[Carlo Lazzaro]

Melanie:
as an aside to Rich's helpful advice, you may want to consider -statsby-:

 use "C:\Program Files\Stata17\ado\base\a\auto.dta"
(1978 automobile data)

. sysuse auto
(1978 automobile data)
. statsby, by(foreign): regress c.mpg##c.gear turn
(running regress on estimation sample)

Command: regress c.mpg##c.gear turn
By: foreign

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

. mat list e(b)

e(b)[1,4]
c.mpg#                            
gear_ratio  c.gear_ratio          turn         _cons
y1    -6.1057963     .27334161    -.09069269       25.4113

.

Usually, when I code an interaction, I ask Stata to provide me with the -e(b)- matrix to double-check the names of the resulting coefficients provided by -statsby-.

[Joro Kolev]

Coefficients are more efficiently stored as scalars, locals, or matrices, rather than variables.

To see how your coefficients are called, you can do the following:

. sysuse auto, clear
(1978 automobile data)

. reg price i.rep##c.mpg, coeflegend

      Source |       SS           df       MS      Number of obs   =        69
-------------+----------------------------------   F(9, 59)        =      3.65
       Model |   206362465         9  22929162.7   Prob > F        =    0.0011
    Residual |   370434494        59  6278550.75   R-squared       =    0.3578
-------------+----------------------------------   Adj R-squared   =    0.2598
       Total |   576796959        68  8482308.22   Root MSE        =    2505.7

------------------------------------------------------------------------------
       price | Coefficient  Legend
-------------+----------------------------------------------------------------
       rep78 |
          2  |   10881.12  _b[2.rep78]
          3  |   8973.281  _b[3.rep78]
          4  |   651.7399  _b[4.rep78]
          5  |   4363.191  _b[5.rep78]
             |
         mpg |  -123.1667  _b[mpg]
             |
 rep78#c.mpg |
          2  |  -507.6563  _b[2.rep78#c.mpg]
          3  |  -375.7209  _b[3.rep78#c.mpg]
          4  |   43.26329  _b[4.rep78#c.mpg]
          5  |  -81.52802  _b[5.rep78#c.mpg]
             |
       _cons |       7151  _b[_cons]
------------------------------------------------------------------------------

.