Is my interpretation correct?

[Olami Mario]

I want to compute the marginal effect of maternal variables on child nutritional status :

maternal variables - financial empowerment, social empowerment, age group, education, work status and household wealth. all of these variables have more than two categories.

Education has 3 categories : none, primary, secondary+

child nutritional status has 4 categories: normal, stunting, underweight, stunting and underweight

my command is :

svy:mlogit nutristat i.financedec i.wempower i.agegroup i.educca i.work i.wealth,rrr

margins, dydx(educca)

marginsplot

Result:

dy/dx w.r.t. : 1.educca 2.educca
1._predict : Pr(nutristat==0), predict(pr outcome(0))
2._predict : Pr(nutristat==1), predict(pr outcome(1))
3._predict : Pr(nutristat==2), predict(pr outcome(2))
4._predict : Pr(nutristat==3), predict(pr outcome(3))


dy/dx

0 education - base outcome

1 education
predict
1 .0113412
2. - . 0131397
3 .007736
4. - .0059375

2 education
predict
1 .1671671
2. - . 0644669
3 .0039641
4. - .1066643


can I interpret mothers education as : Children of mothers with more than secondary level had a 1.7% predicted probability of having good nutritional status and a 0.11% reduced likelihood of been stunted and underweight.

Thank you

[Clyde Schechter]

I don't know where the numbers in your interpretation are coming from, as none of them actually appear in the output, even if I assume you are doing some rounding.

Here's how I would interpret (some of) these results:

Compared to children of mothers with no education, the children of mothers with primary education, the probability of being normal is .0113412 greater (1.13412 percentage points, if you wish to use that scale), the probability of being stunted is .0131397 lower, the probability of being underweight is 0.007736 greater, and the probability of being both stunted and underweight is .0059375 less.

The remainder of the results can be analogously interpreted as a comparison of children of mothers with secondary education vs children of mothers with no education.

[Olami Mario]

Thank you Clyde