Can mcp command be used to compute marginal effects for continuous variable after fitting a xtmelogit model in Stata?

[smith Jason]

The dataset is here:
insheet using "https://stats.idre.ucla.edu/stat/data/hdp.csv", comma
foreach i of varlist familyhx smokinghx sex cancerstage school {
encode `i', gen(`i'2)
drop `i'
rename `i'2 `i'
}

xtmelogit remission il6 crp i.cancerstage lengthofstay c.pain c.pain#i.cancerstage|| did: , intpoints(10)

To this step, everything seems Ok.

However, when I used the Stata code below, the error messages prompted

mcp pain cancerstage

There was a problem executing -margins-.

The command issued was essentially as follows: margins , at( pain=( 1 2 3 4 5 6 7 8 9) cancerstage=( 1 2 3 4)) r(498);

Does anybody know if mcp command is not allowed to be used after fitting a xtemologit model?

Thank you!

[smith Jason]

Anybody here can help me? Thank you!

[smith Jason]

anybody？

[smith Jason]

feel depressed ….,

[Eddy Simms]

I am sorry I cannot answer for mcp or mcp2, but does

margins cancerstage, at(pain=(1(1)9)) plot

do what you want?

[smith Jason]

Thank you for your response! If the model now is
xtmelogit remission il6 crp i.cancerstage lengthofstay c.pain#i.cancerstage c.pain c.pain#c.pain c.pain#c.pain#i.cancerstage|| did: , intpoints(10)

How can I use margins to compute it?

[Joseph Coveney]

If the model now is
xtmelogit remission il6 crp i.cancerstage lengthofstay c.pain#i.cancerstage c.pain c.pain#c.pain c.pain#c.pain#i.cancerstage|| did: , intpoints(10)

How can I use margins to compute it?

It would be the same as Eddy showed. Adding the quadratic term, so long as you use factor variable notation to do it, wouldn't change the margins statement, because margins is designed to work seamlessly with factor variable notation.

Try this and see for yourself.

version 17.0

clear *

import delimited using "https://stats.idre.ucla.edu/stat/data/hdp.csv", ///
    varnames(1) asdouble

label define CancerStages 1 I 2 II 3 III 4 IV
encode cancerstage, generate(stage) label(CancerStages) noextend
drop cancerstage
rename stage cancerstage

melogit remission c.(il6 crp lengthofstay) c.pain##c.pain##i.cancerstage || did: , nolog

margins cancerstage, at(pain = (1 5 9)) // <= Here

marginsplot , xdimension(cancerstage) plotdimension(pain) ///
    noci title("") ///
    ytitle(Remission Rate) ylabel( , format(%02.1f) angle(horizontal) nogrid) ///
    legend(position(7) ring(0))

exit

[smith Jason]

Thank you!

[smith Jason]

Thank you!For the same variable, If I want to plot the marginal effects of M-SD,M, and M+SD, How can I do it? By the way, c_pain is the centered variable and it replaced the c.pain in the model for ease of interpretation.

[smith Jason]

This is the plot I graphed with mcp2 command.