I have a probably quite simple question which I do not know how to implement on Stata.

The background is that I estimate xtprobit separately for the employment probabilities of migrants from different areas of origin (7 categories). To simplify the model for discussion, let's say I control for education level which ranges from low, middle to high, gender and age. For example, for migrants from origin 2:
xtprobit i.employment i.edu i.female age if area_origin==2, i(id)
margins, predict(pu0) at (age=28 edu=1)
And so on for each area of origin.

I want to compare the marginal probabilities for migrants from each area of origin, at each education level and of each gender to one specific threshold, that is, the probability of migrants from EU (area_origin==1) with low education level (edu==1) by running a statistical test to see whether their employment probabilities will be significantly higher than the threshold.

But I have difficulty in how to implement this in practice, as these probabilities come not only from different regressions, I am specifically compare to one group with a specific characteristic (low education level).

Could anyone help in sharing your insights on this? Many, many thanks indeed for your generous help!