Hi all,

I'm trying to get adjusted risk differences (ARD) for several analyses I'm doing with complex survey data, including a few binary logistic regressions and a multinomial logistic regression, and I'm having some issues. I stumbled on the Norton, Miller and Kleinman paper "Computing adjusted risk ratios and risk differences in Stata" in the Stata Journal from 2013, describing their ADJRR command, which seems to work fine but only for binary predictors.

I also found Clyde Schechter's kind answer on here, but couldn't figure out how to replicate it with a command such as:

svy linearized : logit treatment age gender race

...which is trying to predict whether treatment is provided (binary) adjusting for age (continuous), gender (binary, for these purposes) and race (4 categories, for these purposes).

If anyone figured out how to get ARD for covariates which aren't binary using ADJRR, or has any other way to get ARD for binomial or multinomial logistic regressions on svy data that would be greatly appreciated!

Thanks,

John

I'm trying to get adjusted risk differences (ARD) for several analyses I'm doing with complex survey data, including a few binary logistic regressions and a multinomial logistic regression, and I'm having some issues. I stumbled on the Norton, Miller and Kleinman paper "Computing adjusted risk ratios and risk differences in Stata" in the Stata Journal from 2013, describing their ADJRR command, which seems to work fine but only for binary predictors.

I also found Clyde Schechter's kind answer on here, but couldn't figure out how to replicate it with a command such as:

svy linearized : logit treatment age gender race

...which is trying to predict whether treatment is provided (binary) adjusting for age (continuous), gender (binary, for these purposes) and race (4 categories, for these purposes).

If anyone figured out how to get ARD for covariates which aren't binary using ADJRR, or has any other way to get ARD for binomial or multinomial logistic regressions on svy data that would be greatly appreciated!

Thanks,

John

## Comment