Marginal effects and dyadic logit in STATA

[Faradj Koliev]

Dear all,

My question is whether marginal effects make sense when the data is dyadic. Actually, I have no reason to assume anything else, but after asking some colleagues I feel that I want to be sure. I've seen some studies using dyadic data but none have used marginal effects. If you know better than me, please let me know if I miss anything.

This is my model: l

logit dv x1 i.x2 x3 i.x4 i.year, vce(cl IDdyad)

to calculate the average marginal effects I do the following (as I use to do with my logistic models) :

mchange, stats(se, p)

Best,
Faradj

[Richard Williams]

My guess is that marginal effects and adjusted predictions are used as much or more with logistic regression than they are with other methods. In any event they are certainly ok.

You may want to see

https://www3.nd.edu/~rwilliam/stats3/Margins01.pdf

Other related handouts are at

https://www3.nd.edu/~rwilliam/stats3/index.html

Also, you should note where you got mchange. I assume it was part of the spost13 package of Long and Freese.