BGGM (version 2.1.5)

pcor_to_cor: Compute Correlations from the Partial Correlations

Description

Convert the partial correlation matrices into correlation matrices. To our knowledge, this is the only Bayesian implementation in R that can estiamte Pearson's, tetrachoric (binary), polychoric (ordinal with more than two cateogries), and rank based correlation coefficients.

Usage

pcor_to_cor(object, iter = NULL)

Value

• R An array including the correlation matrices (of dimensions p by p by iter)

• R_mean Posterior mean of the correlations (of dimensions p by p)

Arguments

object

An object of class estimate or explore

iter

numeric. How many iterations (i.e., posterior samples) should be used ? The default uses all of the samples, but note that this can take a long time with large matrices.

Examples

# \donttest{
# note: iter = 250 for demonstrative purposes

# data
Y <- BGGM::ptsd

#########################
###### continuous #######
#########################

# estimate the model
fit <- estimate(Y, iter = 250,
                progress = FALSE)

# compute correlations
cors <- pcor_to_cor(fit)


#########################
###### ordinal  #########
#########################

# first level must be 1 !
Y <- Y + 1

# estimate the model
fit <- estimate(Y, type =  "ordinal",
                iter = 250,
                progress = FALSE)

# compute correlations
cors <- pcor_to_cor(fit)


#########################
#######   mixed    ######
#########################

# rank based correlations

# estimate the model
fit <- estimate(Y, type =  "mixed",
                iter = 250,
                progress = FALSE)

# compute correlations
cors <- pcor_to_cor(fit)
# }