Margins at the median and all the other variables at zero level

[Linda Luciani]

Hello people,

I have run my LOGIT regression and now I want to obtain the marginal effect at the median for each single variable, while all the others shall be equal to zero. The following code in yellow and output do it that , but keep the other variables at their median level. How can I play around the code so to obtain what the desired outcome initially described?

Many Thanks!
Best
Linda

--------------------------------------------------------------------------------

[Andrew Musau]

You can explicitly specify a value at which to set a particular variable. However, in your case, some values are not even possible. For example, I am certain that there is no year=0 if your median year is 2019, and I doubt that you have newborns sampled in your study.

webuse lbw, clear
logit low age smoke ptl ht ui
qui sum age, d
local medage= r(p50)
margins, dydx(*) at(age= `medage' smoke==0 ptl==0 ht==0 ui==0)

Res.:

. margins, dydx(*) at(age= `medage' smoke==0 ptl==0 ht==0 ui==0)

Conditional marginal effects                    Number of obs     =        189
Model VCE    : OIM

Expression   : Pr(low), predict()
dy/dx w.r.t. : age smoke ptl ht ui
at           : age             =          23
               smoke           =           0
               ptl             =           0
               ht              =           0
               ui              =           0

------------------------------------------------------------------------------
             |            Delta-method
             |      dy/dx   Std. Err.      z    P>|z|     [95% Conf. Interval]
-------------+----------------------------------------------------------------
         age |  -.0085216   .0054285    -1.57   0.116    -.0191613    .0021181
       smoke |   .0876545    .047171     1.86   0.063    -.0047991     .180108
         ptl |   .1069912   .0537701     1.99   0.047     .0016037    .2123787
          ht |   .2217971   .0953514     2.33   0.020     .0349118    .4086824
          ui |   .1286365   .0678114     1.90   0.058    -.0042715    .2615444
------------------------------------------------------------------------------

[Linda Luciani]

Hi Andrew,

yeah what you say actually makes sense!

Thank you very much for the illustrative example.

Best,
Linda