Proper use of joint/interaction on adjusted predictions (margins) for complex surveys analysis- Juan Albertorio, NCHS

[Juan Albertorio]

Dear StataList,

I have a request. I am applying adjusted predictions (margins) to perform several estimates on the joint effect of two covariates in a qualitative outcome for a complex survey.

I browsed the information that mostly Richard Williams and STATA books, and others (Michael Mitchell) had provided on margins use which is quite excellent. However, a nitty-gritty that they have not cover is to provide relevant evidence of what are the similarities or contrast when you request an interaction or joint effect in margins when using the commands (e.g. #, ##, @, over).

Since I am using complex survey data I rely more on the use of over to address properly the standard error calculation for subpopulation (e.g., male vs. female) but I wonder if I am doing it right and whether I can find documentation that I can rely on for my judgment.

Any idea, suggestion, or criticism is welcome.



Sincerely yours.
Juan

[Clyde Schechter]

Your question is very broad and general. I think your chance of getting a timely and helpful response would be improved by providing a specific example and showing actual commands and output that you have questions about.