Hello,
Thank you so much for this great package!
I am using cmest to estimate direct and indirect effects for a binary outcome, binary exposure, and binary mediator with closed-form parameter function estimation, and my analyses currently employ multiple imputation with chained equations (MICE). I recently read that including interaction terms within MICE can improve the coverage of effect size confidence intervals (please see here), but have not been able to implement this within cmest because there is no option to specify custom interaction terms (similar to options to specify exposures, mediators, and outcomes). I realise that this would be enabled with direct counterfactual imputation estimation, but would prefer an option enabled within closed-form parameter function estimation because this would render a sensitivity analysis that incorporates interaction terms into MICE comparable with my other analyses.
Thank you again, and best wishes,
Yaning
Hello,
Thank you so much for this great package!
I am using
cmestto estimate direct and indirect effects for a binary outcome, binary exposure, and binary mediator with closed-form parameter function estimation, and my analyses currently employ multiple imputation with chained equations (MICE). I recently read that including interaction terms within MICE can improve the coverage of effect size confidence intervals (please see here), but have not been able to implement this withincmestbecause there is no option to specify custom interaction terms (similar to options to specify exposures, mediators, and outcomes). I realise that this would be enabled with direct counterfactual imputation estimation, but would prefer an option enabled within closed-form parameter function estimation because this would render a sensitivity analysis that incorporates interaction terms into MICE comparable with my other analyses.Thank you again, and best wishes,
Yaning