Generate Dummy Variables from Multinomial Distribution

[Eric Wilbrandt]

I am currently stuck on how to program a data generation process, I am using stata 15.1:

I have a probabilistic defined dummy variable, which could be defined as 6 different states.

For example - in the first observation I have the following:

[p1 p2 p3 p4 p5 p6]

with each probability as its own defined variable and all six probabilities sum to 1.

I would like to generate 6 dummy variables from these probabilities:

[d1 d2 d3 d4 d5 d6]

where the probability of the dummy variable equaling 1 is equal to the probabilities defined above. I've tried a binomial distribution, but because these are dependent (all the other dummies must equal 0 if one takes the value of 1) I am having trouble. A multinomial distribution seems likes the best choice, but I am unsure on how to program the generation of such variables without turning to matlab or R.

Thank you for the guidance,

Eric Wilbrandt

[Weiwen Ng]

Mike Lacy gets a tip of the hat for this advice using the -irecode- function, and I think you would type:

gen dummy =  irecode(runiform(),p1,p1+p2,p1+p2+p3,...)
tabulate dummy, gen(d)

Does that work? If that's not quite what you had in mind, can you show a data example?

[Eric Wilbrandt]

Weiwen Ng , this is exactly what I was looking for. Thank you very much for pointing me in the right direction. Mike Lacy thanks for pointing this out in another post.