Ordered Probit: discontinuous region with missing values encountered when estimating Marginal effect

[Neeraj Kumar]

Dear Statalist

I am using ordered probit model. While calculating the marginal effect it showing the error discontinuous region with missing value. My dependent variable is having 10 categories. I have used the following command

xtoprobit A170 LnNSDPpcC X003 X011 marriage employ gender dharm1 dharm2 dharm3  health2 health3 edu2 edu3 edu4 class1 class2 So2 No2 BOD formalscindex2 formalscindex3 informalSc2 informalSc3 , vce(robust)

 margins, dydx(*)

Kindly help me.

[FernandoRios]

Most of the time this is not a problem that can be "fixed"
Simply your model is very complex (many covariates x panel structure)
Panel structure model introduces a level of complexity in the estimation that makes it slow, and often unfeasible.
Possible Solutions:
re consider your model. Do you really need panel? what about something similar to correlated random effects model?
Do you need all explanatory variables?
HTH

[Neeraj Kumar]

Thank you so much Mr. FernandoRios for your reply. I have run the model with different permutation and combination and I have found this particular error is coming when I am including No2 variable. Actually it's panel study. No2 is a continuous variable and one of the important variable for my study. I have deleted all the missing value but if I am including this No2 variable the error is persisting. Actually I need all the variables for this particular study. Some other purpose when I run regoporb2 command with same set of variables and after that when I am running a Margin command that time it giving marginal effect. But y this problem is coming in the case of xtoprobit. Is there any way we can sort out this issue or dropping this variable is the only choice left for me. I don't know anything about correlated random effects model. or if u can guide me on correlated random effect model. How to use in the case of ordial dependent variable with panel data.