Including Marginal Effects in a Table

[Charles Ehmat]

Dear all,

I am running probit regressions to try and find the independent variables that influence my dependent variable. This means running multiple slightly different probit regressions (around 12). For better comparison on which model fits best I want to create a table including all the different models. To do this I am doing the following for each model:

probit res_ov fis_bal fis_exp panel pub_type inc_class start freq_an dep_var_class
estimates store m1, title(Model 1)

probit res_ov fis_bal fis_exp panel pub_type pub_class pub_year
estimates store m2, title(Model 2)

and so on.

estout m1 m2

works perfectly fine to create my table. However I am not interested in the coefficients but rather in the marginal effects of each variable. Hence, for each regression I also run

margins, dydx(*) atmeans

I tried storing them with using estpost margins, dydx(*) atmeans instead and then including them by specifying the table command with the cells option:

estout m1 m2, cells(margins(star fmt(3)) se(par))


As a results I no longer get the coefficients but rather an empty space where I want my marginal effects to be. This did not change when I replaced margins in the last equation with cmd, dydx or any other possible name for the margins.



Additionally I have a question about the table stata is giving me. Up until estout m1 m2 m3 m4 the table looks perfectly fine. But if I include more models the table puts those in a second row which makes evaluating the table very difficult. Is there a command that tells stata to put all models next to each other?


Thanks already for your help! I greatly appreciate it.

Charles



(I am using Stata on different computers with versions from 14 to 16 both on Windows and on Mac )

[Phil Bromiley]

You didn't get a quick answer. You'll increase your chances of a useful answer by following the FAQ on asking questions - provide Stata code in code delimiters (which you do), readable Stata output, and sample data using dataex.

If you put post on the margins statement, it will save a bunch of stuff - look at it by issuing return list and ereturn list at the command line. You can work with these - put them into table or matrices or whatever. There is probably a way to trick outreg2 or another program to use them in the output, but getting it really neat is hard.

Outreg2 and other procedures will not wrap the output as apparently estout does.