CDE_cont: Controlled Direct Effect (CDE) function for cases when...

In harvard-P01/causalMediation: Causal Mediation Analysis

Description

This function calculates the Controlled Direct Effect (CDE) for continuous outcomes as described in the publication of Valeri and VanderWeele (2013, doi: 10.1037/a0031034).

Usage

CDE_cont(thetas, treatment, mediator, m = 0, a_old = 1, a_new = 0)

Arguments

thetas	The coefficients of the regression on the outcome variable.
treatment	The column name in the dataframe which contains the treatment values.
mediator	The column name in the dataframe which contains the mediator values.
m	The value of the mediator at which to evaluate the result. Default = 0.
a_old	The old value of the treatment at which to evaluate the result (a in the publication of Valeri and VanderWeele). Default = 1.
a_new	The new value of the treatment at which to evaluate the result (a* in the publication of Valeri and VanderWeele). Default = 0.

Note

This package is under continuous (but perhaps slow) development. Users are welcome to contribute to the development.

Author(s)

Egge van der Poel ([e-mail]).

References

Valeri and VanderWeele (2013, doi: 10.1037/a0031034).

Examples

df <- data.frame('smoking'    = c(0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 0),
                 'lbw'        = c(0, 0, 0, 1, 0, 0, 0, 1, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0),
                 'death'      = c(0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0),
                 'drinking'   = c(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0),
                 'agebelow20' = c(0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0)
                )

betas  <- c('(Intercept)' = -2.4916567, 'smoking' = 0.4905681, 'drinking' = -0.5967808, 
            'agebelow20' = 0.2305012)
thetas <- c('(Intercept)' = -6.5108483, 'smoking' = 0.5290551, 'lbw' = 3.5778582, 
            'drinking' = -0.3125368, 'agebelow20' = 0.2668849, 'smoking:lbw' = -0.5336013)
outcome='death'
treatment='smoking'
mediator='lbw'
covariates=c('drinking', 'agebelow20')

CDE_cont(thetas = thetas, treatment = treatment, mediator = mediator)

## The function is currently defined as
function (thetas, treatment, mediator, m = 0, a_old = 1, a_new = 0) 
{
    interactionTerm <- ifelse(is.na(thetas[paste(treatment, mediator, 
        sep = ":")]), 0, thetas[paste(treatment, mediator, sep = ":")])
    cde <- (thetas[treatment] + interactionTerm * m) * (a_old - 
        a_new)
    unname(cde)
  }

harvard-P01/causalMediation documentation built on May 17, 2019, 3:04 p.m.