[Sidika Tunc Candogan]

Dear all, I have a question about marginsplot after bootstrapping:
Here is my two-stage model:

*** IV Model 1
capture program drop boot
program boot, rclass
***First Stage***
reg X i.IV Controls, robust
predict Residuals, resid
est store Residuals

***Second Stage***
probit Y c.X##c.X c.Residuals##c.Residuals Controls
margins, at(X=(0(1)48)) post
est store margins3
drop Residuals
end
bootstrap _b, reps(3) seed(123): boot

Then, I use the following command to plot margins:
est restore margins3
margins
marginsplot

I can see margins and their confidence intervals, but I am getting the following error: "variable Residuals not found" Margins and their confidence interval are already calculated, why does marginsplot command need Residuals? Is there any other command to plot already calculated margins and their confidence intervals?

Thank you very much!

[FernandoRios]

Hi Sidika
The reason for the error is because how margins (and marginsplot) work. It is trying to verify ALL variables in the original model exist, even if they are not part of the plot of interest.
The easiest solution is just to create your variable "residuals" equal to 0. and then make the plot.
HTH
Fernando

[Sidika Tunc Candogan]

FernandoRios Dear Fernando,

Thank you very much for your response! It worked perfectly, I have been trying to solve this problem for a long time, appreciate your help a lot!