i have combined survey data sets from 32 countries and have about 90k respondents, the number of respondents doesn't correlate to countries populations, I am looking for global factors associated with suicide ideation (0 no or 1 yes)

i am looking at 18 different variables, but a few questions were omitted from a country's survey and also there are randomly missing responses of the respondent just didn't answer.

is MI chained okay to deal with all the missing data (most questions are missing less than 10% and only 2 have between 10-20% missing data due to some countries omitting these questions)

my second issue is trying to figure out how to weight the respondents differently based on the differences in population for each country and the number of respondents. is there some way to weight respondents 1-3000 as something, 3001-6000 as another weight, and so on. or should I even do this? I am looking for global factors associated with suicidal ideation, I don't care about country or region specific.

Thank you to anyone who can help me out, I am very new to stata and it feels like learning a new language hah