Comparing multiple proportions in Stata

[May Cher]

Comparing multiple proportions in Stata;

What is the command to compare >3 proportions in Stata?

[Mike Lacy]

Do you mean that you have a single binary outcome, to be compared across > 3 groups of individuals, so that you are interested in how the proportion of "yes" varies across those three groups? In that case, two possibilities are:

tab2 outcome group, col
logit outcome, i.group

where "outcome" is your binary response variable, and "group" is the explanatory variable with > 3 categories.

[May Cher]

Thank you. No, that is not what I am looking for. I have proportions and not raw data, so I don't have the real variables. I am interested in comparing different proportions (0.15; 0.4; 0.9, ...) to another major proportion and see if there is a significant difference between the proportions, and if so, between which ones. It is as if I am doing Chi-square for multiple [proportions. Is there a command in Stata for that? I also read online that I should do Marascuilo Method to compare multiple proportions. Is there a command in stata to run that Marascuilo Method?

[Leonardo Guizzetti]

I'm not familiar with the Marascuilo method, but a quick search turned up this link, which indicates that this method goes beyond an omnibus testing procedure to consider the simultaneous pairwise differences among different populations. If this is what you are looking for, I don't see an existing program that does this, but it could be written.

[May Cher]

Thank you Leonardo Guizzetti. Yes, this is it. How long does it take to get that written in Stata? Can you get a command for it?

[Leonardo Guizzetti]

I'm sorry, no I do not have time to do this myself. Depending on your level of familiarity with Stata you could can implement this yourself, even if ad hoc and not in a neat package. This could be a good learning opportunity for you, but if time is a concern, perhaps another route is needed.