Testing joint significance of multiple coefficients in probit after margins

[Mats Krasinski]

Hi. After probit I am calculating marginal effects via

magins, dydx(*)

.
Then I want to jointly test that covariates are equal to 0 and get chi statistic. How should I do that? I found some parameter called "contrast" but this does not work.

[Joseph Coveney]

After probit . . . I want to jointly test that covariates are equal to 0 and get chi statistic. How should I do that?

It's already right there in the header of the probit regression output.

Look in the upper-right side and you'll see LR chi2(<df>) = <some value>.

That's it.

[Mats Krasinski]

It's already right there in the header of the probit regression output.

Look in the upper-right side and you'll see LR chi2(<df>) = <some value>.

That's it.

Thank you for replying. It is for probit's coefficients but I need for coefficients of marginal effect.

[Richard Williams]

Does this give you what you want?

webuse nhanes2f, clear
probit diabetes i.black i.sex height weight, nolog
margins, dydx(*) post
testparm i.black i.sex height weight

. webuse nhanes2f, clear

. probit diabetes i.black i.sex height weight, nolog

Probit regression                                       Number of obs = 10,335
                                                        LR chi2(4)    = 152.46
                                                        Prob > chi2   = 0.0000
Log likelihood = -1922.8344                             Pseudo R2     = 0.0381

------------------------------------------------------------------------------
    diabetes | Coefficient  Std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
       black |
      Black  |   .2390503   .0617771     3.87   0.000     .1179695    .3601311
             |
         sex |
     Female  |  -.1545867    .061044    -2.53   0.011    -.2742308   -.0349426
      height |  -.0278796   .0033436    -8.34   0.000    -.0344328   -.0213263
      weight |   .0141805   .0014354     9.88   0.000     .0113671    .0169938
       _cons |   1.996139   .5546783     3.60   0.000     .9089894    3.083289
------------------------------------------------------------------------------

. margins, dydx(*) post

Average marginal effects                                Number of obs = 10,335
Model VCE: OIM

Expression: Pr(diabetes), predict()
dy/dx wrt:  1.black 2.sex height weight

------------------------------------------------------------------------------
             |            Delta-method
             |      dy/dx   std. err.      z    P>|z|     [95% conf. interval]
-------------+----------------------------------------------------------------
       black |
      Black  |   .0267878   .0079284     3.38   0.001     .0112484    .0423272
             |
         sex |
     Female  |  -.0152008   .0061195    -2.48   0.013    -.0271948   -.0032067
      height |  -.0026978   .0003313    -8.14   0.000    -.0033471   -.0020485
      weight |   .0013722   .0001433     9.58   0.000     .0010914     .001653
------------------------------------------------------------------------------
Note: dy/dx for factor levels is the discrete change from the base level.

. testparm i.black i.sex height weight

 ( 1)  1.black = 0
 ( 2)  2.sex = 0
 ( 3)  height = 0
 ( 4)  weight = 0

           chi2(  4) =  136.19
         Prob > chi2 =    0.0000

.

[Mats Krasinski]

Does this give you what you want?

webuse nhanes2f, clear
probit diabetes i.black i.sex height weight, nolog
margins, dydx(*) post
testparm i.black i.sex height weight

. webuse nhanes2f, clear

. probit diabetes i.black i.sex height weight, nolog

Probit regression Number of obs = 10,335
LR chi2(4) = 152.46
Prob > chi2 = 0.0000
Log likelihood = -1922.8344 Pseudo R2 = 0.0381

------------------------------------------------------------------------------
diabetes | Coefficient Std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
black |
Black | .2390503 .0617771 3.87 0.000 .1179695 .3601311
|
sex |
Female | -.1545867 .061044 -2.53 0.011 -.2742308 -.0349426
height | -.0278796 .0033436 -8.34 0.000 -.0344328 -.0213263
weight | .0141805 .0014354 9.88 0.000 .0113671 .0169938
_cons | 1.996139 .5546783 3.60 0.000 .9089894 3.083289
------------------------------------------------------------------------------

. margins, dydx(*) post

Average marginal effects Number of obs = 10,335
Model VCE: OIM

Expression: Pr(diabetes), predict()
dy/dx wrt: 1.black 2.sex height weight

------------------------------------------------------------------------------
| Delta-method
| dy/dx std. err. z P>|z| [95% conf. interval]
-------------+----------------------------------------------------------------
black |
Black | .0267878 .0079284 3.38 0.001 .0112484 .0423272
|
sex |
Female | -.0152008 .0061195 -2.48 0.013 -.0271948 -.0032067
height | -.0026978 .0003313 -8.14 0.000 -.0033471 -.0020485
weight | .0013722 .0001433 9.58 0.000 .0010914 .001653
------------------------------------------------------------------------------
Note: dy/dx for factor levels is the discrete change from the base level.

. testparm i.black i.sex height weight

( 1) 1.black = 0
( 2) 2.sex = 0
( 3) height = 0
( 4) weight = 0

chi2( 4) = 136.19
Prob > chi2 = 0.0000

.

thank you very much. that's what I need.