xmed function

Function to performed exploratory mediation with continuous and categorical variables

xmed(
  data,
  iv,
  mediators,
  dv,
  covariates = NULL,
  type = "lasso",
  nfolds = 10,
  show.lambda = F,
  epsilon = 0.001,
  seed = NULL
)

Arguments

• data: Name of the dataset
• iv: Name (or vector of names) of independent variable(s)
• mediators: Name of mediators
• dv: Name of dependent variable
• covariates: Name of covariates to be included in model.
• type: What type of penalty. Options include lasso, ridge, and enet.
• nfolds: Number of cross-validation folds.
• show.lambda: Displays lambda values in output
• epsilon: Threshold for determining whether effect is 0 or not.
• seed: Set seed to control CV results

Returns

Coefficients from best fitting model

Examples

# example
library(ISLR)
College1 = College[which(College$Private=="Yes"),]
Data = data.frame(scale(College1[c("Grad.Rate","Accept","Outstate","Room.Board","Books","Expend")]))
Data$Grad.Rate <- ifelse(Data$Grad.Rate > 0,1,0)
Data$Grad.Rate <- as.factor(Data$Grad.Rate)
#lavaan model with all mediators
model1 <-
 ' # direct effect (c_prime)
Grad.Rate ~ c_prime*Accept
# mediators
Outstate ~ a1*Accept
Room.Board ~ a2*Accept
Books ~ a3*Accept
Expend ~ a6*Accept
Grad.Rate ~ b1*Outstate + b2*Room.Board + b3*Books + b6*Expend
# indirect effects (a*b)
a1b1 := a1*b1
a2b2 := a2*b2
a3b3 := a3*b3
a6b6 := a6*b6
# total effect (c)
c := c_prime + (a1*b1) + (a2*b2) + (a3*b3) + (a6*b6)
'
#p-value approach using delta method standard errors
fit.delta = sem(model1,data=Data,fixed.x=TRUE,ordered="Grad.Rate")
summary(fit.delta)

#xmed()

iv <- "Accept"
dv <- "Grad.Rate"
mediators <- c("Outstate","Room.Board","Books","Expend")

out <- xmed(Data,iv,mediators,dv)
out