Accessing values returned by -margins-

[Gobinda Natak]

Hi everyone

I'm interested whether a variable (foreign) modifies the effect of X (trunk) on Y (mpg). Here's an example:

sysuse auto, clear

regress mpg c.trunk##foreign

To do so, I test the difference in slopes for domestic and foreign cars with -margins-:

margins, dydx(trunk) over(foreign) pwcompare(effects)

Super easy, yields the following output:

Pairwise comparisons of average marginal effects

Model VCE: OLS                                              Number of obs = 74

Expression: Linear prediction, predict()
dy/dx wrt:  trunk
Over:       foreign

--------------------------------------------------------------------------------------
                     |   Contrast Delta-method    Unadjusted           Unadjusted
                     |      dy/dx   std. err.      t    P>|t|     [95% conf. interval]
---------------------+----------------------------------------------------------------
trunk                |
             foreign |
Foreign vs Domestic  |   .2638674   .3493065     0.76   0.453    -.4328025    .9605373
--------------------------------------------------------------------------------------

My question, as I have to do this one billion times: What is the best practice to access the contrast (0.2638674) and its p-value (0.453) to write it into the notes of figures? E.g. _b[trunk:1vs0.foreign]] doesn't work

Thanks so much
Go

[Andrew Musau]

See

return list

sysuse auto, clear
regress mpg c.trunk##foreign
margins, dydx(trunk) over(foreign) pwcompare(effects)
mat l r(table_vs)
display r(table_vs)["b", 1]
display %4.3f r(table_vs)["pvalue", 1]

Res.:

. margins, dydx(trunk) over(foreign) pwcompare(effects)

Pairwise comparisons of average marginal effects

Model VCE: OLS                                              Number of obs = 74

Expression: Linear prediction, predict()
dy/dx wrt:  trunk
Over:       foreign

--------------------------------------------------------------------------------------
                     |   Contrast Delta-method    Unadjusted           Unadjusted
                     |      dy/dx   std. err.      t    P>|t|     [95% conf. interval]
---------------------+----------------------------------------------------------------
trunk                |
             foreign |
Foreign vs Domestic  |   .2638674   .3493065     0.76   0.453    -.4328025    .9605373
--------------------------------------------------------------------------------------

. 
. mat l r(table_vs)

r(table_vs)[9,1]
             trunk:
              1vs0.
           foreign
     b   .26386741
    se   .34930651
     t   .75540364
pvalue   .45254119
    ll  -.43280246
    ul   .96053728
    df          70
  crit   1.9944371
 eform           0

. 
. display r(table_vs)["b", 1]
.26386741

. 
. display %4.3f r(table_vs)["pvalue", 1]
0.453

.