Calculating margins for regressions which uses plausible values and interactions

[Hannah Elisabeth]

Hi,

I conducted a linear regression with interactions by use of Pisa data.
My code to calculate them is
pv, pv(pv*math) weight(w_fstuwt) brr rw(w_fsturwt*) fays(0.5): ///
reg @pv c.escs##cnt_num i.spktestlang##cnt_num [aw=@w]
escs is the socioeconomic status of an indivdiual
cnt_num is the encoded country (5 countries in the sample)
spktestlang states if the child speaks the language of assessment at home (dummy-variable)

Now, I would like to calculate margins for the effects.

But it does not work and I get an error code instead of a margins table.
More specifically, I have tried different commands like
margins dydx(i.spktestlang c.escs) moptions(atmeans)
just margin and others
But mostly I get this error:
warning: cannot perform check for estimable functions.
e(sample) does not identify the estimation sample

I have also tried to include if e(sample) but it does not work either.

Since Pisadata is a common dataset, I hope that you can help me.
That would be great.

Best,
Hannah

[George Ford]

have you looked at: search pisatools
pisareg might work better.