heterogene effects in stata

[Rasmus Bang Borup]

Hi everyone

I am currently in a situation where i want to figure out the best ways of measuring heterogene effects using stata. Specifically it is between three variables. the outcome variabel is a measure of welfare preferences and is coded binary (coded 0 1), my two other variabels is measures of age and having children. Age is meausured categorical (0-4) and having children is binary (0 1). What i want to know is if there is any heterogene effects of having children on welfare preferences within each agegroup.

I have tried with interaction terms: logit yvar xvar##zvar and the running a margins commando afterwards. My problem is that this commando is measuring the effect to the reference group, which is not useful for interpreting any heterogene effects.

Alder1 is my age variabel
BørnUnder is my having children variabel

I hope the information i provided is enough so that you can help me

[Carlo Lazzaro]

Rasmus:
have you already considered something like -suest-?

[Rasmus Bang Borup]

I have considered this. what i want is to estimate the AME - average marginal effect for having children for each ages group on political preferences. when i run the margins comando after: margins, dydx(BørnUnder) at(Alder1=( 0 1 2 3 4) vsquish, i think i estimate the AME, but im not really sure