Generation of a random dc

[Wali Amar]

Hi,

For a class exercise to show an association between the exposure and the outcome, I want to create a random dichotomous variable (0,1) in a real dataset based on another real dichotomous variable. Is there a way to create such a variable?

Many thanks!

[Joseph Coveney]

Maybe something like the following.

set seed 43736551
assert inlist(real_dichotomous, 0, 1) | mi(real_dichotomous)
generate byte random_dichotomous = ///
    binomial(1, cond(real_dichotomous == 1, 0.75, 0.25)) if !mi(real_dichotomous)

You can adjust the conditional probabilities in order to strengthen or lessen the degree of association.

[Wali Amar]

Hi, thanks. I tried running this syntax but it doesn't run. Stata says it is invalid.

[Joseph Coveney]

Missed an important letter. Sorry about that.

Try this instead.

set seed 43736551
assert inlist(real_dichotomous, 0, 1) | mi(real_dichotomous)
generate byte random_dichotomous = ///
    rbinomial(1, cond(real_dichotomous == 1, 0.75, 0.25)) if !mi(real_dichotomous)

[Wali Amar]

Thanks Joseph. Could you kindly let me know how can I lessen the association? Right now, it shows an OR of 10.9.

Also, I want to see if some other real covariates such as age, and education are confounders or not in this association. Can you tell me how I can look at that?