piieffect: piieffect

In isabelfulcher/frontdoorpiie: Population Intervention Indirect Effect estimation

Description

This function provides estimation and inference for the Population Intervention Indirect Effect (PIIE) as described in Fulcher et al. (2017). The outcome and intermediate variables must be continuous as a linear model is used to model the means. Similarly, the exposure variable must be binary as a logistic model is used.

Usage

piieffect(data, outcome, intermediate, exposure, covariates.outcome,
  covariates.intermediate, covariates.exposure, interaction, astar)


  ## S4 method for signature 
## 'data.frame,
##   character,
##   character,
##   character,
##   vector,
##   vector,
##   vector,
##   numeric,
##   numeric'
piieffect(data,
  outcome, intermediate, exposure, covariates.outcome = c(1),
  covariates.intermediate = c(1), covariates.exposure = c(1),
  interaction = 1, astar = 0)

Arguments

data	A dataframe
outcome	The variable name for the outcome variable in data
intermediate	The variable name for the intermediate variable in data
exposure	The variable name for the exposure variable in data
covariates.outcome	A vector of variable names for covariates to be included in the outcome model
covariates.intermediate	A vector of variable names for covariates to be included in the outcome model
covariates.exposure	A vector of variable names for covariates to be included in the exposure model
interaction	A binary variable indicating if an interaction term between intermediate and exposure is needed
astar	A numeric value for the level of the exposure the intermediate value takes

Methods (by class)

• data = data.frame,outcome = character,intermediate = character,exposure = character,covariates.outcome = vector,covariates.intermediate = vector,covariates.exposure = vector,interaction = numeric,astar = numeric: Generic/Function

Author(s)

Isabel Fulcher

Examples

#Load example dataset
simdata <- readRDS(system.file("rds","simdata1.rds",package="frontdoorpiie"))
#Create an interaction term among covariates
simdata$c1c2 <- simdata$c1*simdata$c2
#Apply the function to estimate PIIE
example <- piieffect(data=simdata,outcome="y",intermediate="m",exposure="a",
covariates.outcome=c("c1","c2","c1c2"),covariates.intermediate=c("c1","c2","c1c2"),covariates.exposure=c("c1","c2","c1c2"),
interaction=1,astar=0)

isabelfulcher/frontdoorpiie documentation built on May 28, 2019, 9:56 p.m.