Calculating risk ratios and risk difference following multiple imputation.

[George Freer]

Hi all,
I have performed multiple imputation by chained equations (MICE) for the dataset I am working with. Is there a way of calculating risk ratios/risk difference using the 'mi estimate' command. I've only been able to work out ORs and coefficients so far. My outcome in binary, so I've tried using both logistic regression and binomial regression.
Would really appreciate any help.

[Rich Goldstein]

-mi estimate- works with -poisson- (risk ratios; make sure you use robust SEs) and -regress- (risk differences) so I would try those

[George Freer]

Thanks for the help. Unfortunately it is the same issue that the only output is a coefficient.

[Rich Goldstein]

I guess I don't understand at all - the coefficient with Poisson, if you use the IRR option, is a risk ratio when the outcome is binary; similarly, the coefficient from -regress- is a risk difference when the outcome is binary

are you saying that these are not what you want? if yes, I don't understand and either you have to try again or maybe someone else can help