Using margins after meglm - How to get the marginal effects after a multi-level model estimated with meglm (works with logit but not meglm)?

[Johannes Mueller]

Dear all,

I want to obtain the marginal effects after estimating a (logit) multi-level model using the meglm command.

The problem I am facing is that I receive the following error message when estimating the multi-level model but NOT when using a (single-level) logit with cluster robust standard errors (clustered at the higher level2)

could not calculate numerical derivatives -- discontinuous region with missing values
encountered
r(459);

While this works...

logit Z $controls c.A##c.B  , vce(cluster HigherLevel) or      
quietly margins, at(A =(-0.9(1)3.2) B=(-1.2(1)3.2)) saving(predictions1, replace)

...this doesn't:

meglm Z $controls c.A##c.B   || HigherLevel: ,  family(binomial) link(logit)  intpoints(14) or
quietly margins, at(A =(-0.9(1)3.2) B=(-1.2(1)3.2)) saving(predictions2, replace)

The model is specified correctly, so I am suspecting it is related to the random part of the MLM, but I am not sure what to do about or where exactly the problem is rooted.
I would be greatful for any hints!!

Thank you in advance!