Tag: sem

Tools

R

R packages: lavaan, semTools

Instructions

• We will test the moderating effect of mod on the two effects of iv1 and iv2 on dv
• Conduct the double mean centering procedure for interactions

Load packages

library(lavaan)
library(semTools)

Measurement model

measure.model <- '
# independent variable 1
iv1 =~ x1 + x2  
# independent variable 2
iv2 =~ x3 + x4 
# moderation variable
mod =~ mod1 + mod2 
# dependent variable
dv =~ y1 + y2 + y3 
'

fit <- sem(measure.model, data = data)
summary(fit, standardized = T)
# standardizedSolution(fit)

Double mean centering

# interaction between iv1 and mod
data.dmc1 <- indProd(data,
var1 = c("x1", "x2"),
var2 = c("mod1", "mod2"),
match = F, meanC = T, residualC = T, doubleMC = T)
# generate new columns: x1.mod1, x2.mod1, x1.mod2, x2.mod2

# interaction between iv2 and mod
data.dmc2 <- indProd(data.dmc1,
var1 = c("x3", "x4"),
var2 = c("mod1", "mod2"),
match = F, meanC = T, residualC = T, doubleMC = T)
# generate new columns: x3.mod1, x4.mod1, x3.mod2, x4.mod2

Moderation model

mod.model <- '
# independent variable 1
iv1 =~ x1 + x2  
# independent variable 2
iv2 =~ x3 + x4 
# moderation variable
mod =~ mod1 + mod2 
# dependent variable
dv =~ y1 + y2 + y3 

# interaction between iv1 and mod
int1 = ~ x1.mod1 + x2.mod1 + x1.mod2 + x2.mod2

# interaction between iv2 and mod
int2 = ~ x3.mod1 + x4.mod1 + x3.mod2 + x4.mod2

# sem model
dv ~ iv1 + iv2 + mod + int1 + int2
'

# use the double mean centered data
fit <- sem(mod.model, data = data.dmc2) 
summary(fit, standardized = T)
# standardizedSolution(fit)