Learn R Programming

metaSEM (version 1.5.0)

Mathieu15: Correlation Matrices from Mathieu et al. (2015)

Description

The data set includes a list of correlation matrices of panel studies between cohesion (C) and performance (P) in Mathieu et al. (2015, Table 1).

Usage

data(Mathieu15)

Arguments

Details

A list of data with the following structure:

data

A list of studies of correlation matrices. The variables are C1, P1, C2, and P2.

n

A vector of sample sizes.

Year

Year of publication.

Sample

Sample characteristics.

Student

Whether the samples are student or non-student based on Sample.

Examples

Run this code
# \donttest{
# TSSEM
## Model 1: no constraint
## Stage 1 analysis
tssem1.fit <- tssem1(Mathieu15$data, Mathieu15$n)
summary(tssem1.fit)

## Proposed model in lavaan syntax
model1 <- 'C2 ~ c2c*C1 + p2c*P1
           P2 ~ c2p*C1 + p2p*P1
           C1 ~~ c1withp1*P1
           C1 ~~ 1*C1
           P1 ~~ 1*P1
           C2 ~~ c2withp2*P2'

## Convert the lavaan model to RAM specification
RAM1 <- lavaan2RAM(model1, obs.variables=c("C1", "P1", "C2", "P2"))
RAM1

## Stage 2 analysis
tssem1b.fit <- tssem2(tssem1.fit, RAM=RAM1)
summary(tssem1b.fit)

plot(tssem1b.fit, col="yellow", edge.label.position=0.58)

## Model 2: Equality constraints on the path coefficient
## Proposed model with equal effects time 1 to time 2
model2 <- 'C2 ~ same*C1 + diff*P1
           P2 ~ diff*C1 + same*P1
           C1 ~~ c1withp1*P1
           C1 ~~ 1*C1
           P1 ~~ 1*P1
           C2 ~~ c2withp2*P2'

## Convert the lavaan model to RAM specification
RAM2 <- lavaan2RAM(model2, obs.variables=c("C1", "P1", "C2", "P2"))
RAM2
    
## Stage 2 analysis
tssem2b.fit <- tssem2(tssem1.fit, RAM=RAM2)
summary(tssem2b.fit)

## Compare the models with and without the constraints. 
anova(tssem1b.fit, tssem2b.fit)

## Plot the model
plot(tssem2b.fit, col="yellow", edge.label.position=0.60)


## OSMASEM
my.df <- Cor2DataFrame(Mathieu15)
    
head(my.df$data)

## Model without any moderator
osmasem.fit1 <- osmasem(model.name="No moderator", RAM=RAM1, data=my.df)
summary(osmasem.fit1)

## Extract the heterogeneity variance-covariance matrix
diag(VarCorr(osmasem.fit1))

plot(osmasem.fit1, col="yellow", edge.label.position=0.6)

## Model with student sample as a moderator on the regression coefficients
A1 <- create.modMatrix(RAM1, output="A", "Student")
A1

## Model with a moderator    
osmasem.fit2 <- osmasem(model.name="Student sample as a moderator", RAM=RAM1, 
                        Ax=A1, data=my.df)
summary(osmasem.fit2)

## Compare the models with and without the moderator
anova(osmasem.fit2, osmasem.fit1)

## Get the R2 of the moderator
osmasemR2(osmasem.fit2, osmasem.fit1)
# }

Run the code above in your browser using DataLab