Dear Stata Users,

I read about the procedure of Goldstein and Healy (1995) to graph the confidence intervals of the predicted probabilities in this post:

https://statisticsbyjim.com/hypothes...compare-means/

According to this post and the original paper of GH, I should use 83% confidence intervals for each group mean
to show that non-overlapping confidence intervals are significantly different at the 95% level.

From my understanding, this could be done in the following straightforward way:

margins x, level(83)

I would like to know whether it is correct or if the CIs should instead be computed in other ways. And I would like to know if you would recommend using this technique for the representation of the marginal effects/predicted probabilities.

Thank you very much
Lydia