adjusted difference in proportions

[Anesh Panchal]

how do i do an adjust one sided normal z test for difference in proportions currently I am using

prtest Primary, by(Trt) level(80) but I want to adjust for one variable while doing this

would it be valid if I did a logistic regression ... logit primary trt var1, or

then halving the p value as it is a one sided test

[Rich Goldstein]

if you use -logit- (or -logistic-) you will lose information; however, there are commands for proportions such as -fracreg- and -betafit- (-betafit- is a user-written program; use -search- to find and install) that can be used

[Joseph Coveney]

You could also consider using the official Stata glm to fit a risk-difference (difference-in-proportions) model.

glm Primary i.Trt var1, family(binomial) link(identity)

You could use a likelihood-ratio test with it

help lrtest