The `mediator` R function conducts mediation analysis under the counterfactual framework assuming interation between the exposure and mediator. Currently the function works for binary and continuous outcomes and mediators.
mediator(...) # S3 method for default mediator( data, out.model, med.model, treat, a = 1, a_star = 0, m = 0, boot_rep = 0, pm_ci = FALSE, ... )
| ... | other arguments |
|---|---|
| data | Data set to use for analysis |
| out.model | A fitted model object for the outcome. Can be of class 'glm','lm', or 'coxph'. |
| med.model | A fitted model object for the mediator. Can be of class 'glm','lm'. |
| treat | A character string indicating the name of the treatment/exposure variable used. |
| a | A numeric value indicating the exposure level. Default = 1 |
| a_star | A numeric value indicating the compared exposure level. Default = 0. |
| m | A numeric value indicating the level of the mediator. Default = 0. |
| boot_rep | A numeric value indicating the number of repetitions to use when utalizing bootstrap to calculate confidence intervals. When `boot_rep` = 0, the Delta method for calculating confidence intervals is used. Default = 0. |
| pm_ci | A logical indicator for calculating the CI for the proportion mediated. Default = FALSE. Currently, the CI can only be determined using boostrapping. If `pm_ci` = TRUE and `boot_rep` = 0 then 100 replicated are automatically used. |
Tibble containing point estimates and 95 percent CI for the CDE, NDE, NIE and TE and the point estimate for the proportion mediated.