Hi,
I am trying to get average marginal effect after xtmlogit.
Here is the example of using -estatus-.
Why do I get different results for hhincome?
The way I understand is that the two commands should give the same results for hhincome, and the second command is a shorthand for getting marginal effects of multiple variables. Do I miss anything?
I am trying to get average marginal effect after xtmlogit.
Here is the example of using -estatus-.
Code:
webuse estatus
xtset id year
xtmlogit estatus i.hhchild##(c.hhincome l2.c.hhincome)
margins hhchild, dydx(hhincome)
----------------------------------------------------------------------------------
| Delta-method
| dy/dx std. err. z P>|z| [95% conf. interval]
-----------------+----------------------------------------------------------------
hhincome |
_predict#hhchild |
1#No | .0003457 .0004927 0.70 0.483 -.00062 .0013113
1#Yes | .0005054 .0004882 1.04 0.301 -.0004514 .0014622
2#No | -.0031088 .0004538 -6.85 0.000 -.0039983 -.0022194
2#Yes | -.0032074 .0003835 -8.36 0.000 -.003959 -.0024558
3#No | .0027631 .0005181 5.33 0.000 .0017477 .0037786
3#Yes | .002702 .0004685 5.77 0.000 .0017838 .0036202
----------------------------------------------------------------------------------
margins hhchild, dydx(hhincome L2.hhincome)
----------------------------------------------------------------------------------
| Delta-method
| dy/dx std. err. z P>|z| [95% conf. interval]
-----------------+----------------------------------------------------------------
hhincome |
_predict#hhchild |
1#No | .0005094 .0027027 0.19 0.851 -.0047878 .0058066
1#Yes | -.0001652 .0024298 -0.07 0.946 -.0049275 .0045971
2#No | -.0021406 .0021466 -1.00 0.319 -.0063479 .0020666
2#Yes | -.0020974 .0017222 -1.22 0.223 -.0054729 .001278
3#No | .0016312 .0029516 0.55 0.580 -.0041539 .0074164
3#Yes | .0022626 .0025153 0.90 0.368 -.0026674 .0071926
-----------------+----------------------------------------------------------------
L2.hhincome |
_predict#hhchild |
1#No | -.0001637 .0027309 -0.06 0.952 -.0055161 .0051887
1#Yes | .0006706 .0024668 0.27 0.786 -.0041643 .0055055
2#No | -.0009682 .0021789 -0.44 0.657 -.0052387 .0033023
2#Yes | -.00111 .0017582 -0.63 0.528 -.004556 .002336
3#No | .0011319 .0029834 0.38 0.704 -.0047154 .0069792
3#Yes | .0004394 .0025543 0.17 0.863 -.0045669 .0054458
----------------------------------------------------------------------------------
The way I understand is that the two commands should give the same results for hhincome, and the second command is a shorthand for getting marginal effects of multiple variables. Do I miss anything?
