correlogram with reliabilities in the main diagonal - r

Question

I'm rather new to R, here is something I encountered in my first steps with it.
In some papers it is required to present a correlogram with the reliability (Cronbach Alpha) of the the correlated variables in the main diagonal (where the correlations are 1 )
an example might be 5 correlated psychometric measures
Job_ins (an average of 4 items)
Employability (an average of 4 items)
INT_to_quit (an average of 4 items)
Mobility_pref (an average of 5 items)
Career_self_mgmt (an average of 8 items)
note that in the native cor() R function the main diagnal (the correlations of the measures with themselves) shows 1.
What I would like to do is to present internal reliablity (cronbach alphas) in the main diagonal instead.
any ideas?
Saar

Answer 1

If I understood you correctly, this is my (long) solution.
#Loading pkgs
require(tidyverse)
require(Hmisc)
require(psych)
#Creating example data
set.seed(123) #making the random data reproducible
#Creating the items for each subject
job <- sim.congeneric(N=200, short = FALSE, low = 1, high=10,categorical=TRUE)
Employability <- sim.congeneric(N=200, short = FALSE, low = 1, high=10,categorical=TRUE)
Career <- sim.congeneric(N=200, short = FALSE, low = 1, high=10,categorical=TRUE)
#Arranging the data to one data.frame
df <- data.frame(as.data.frame(job$observed) %>%
set_names(c("job1", "job2", "job3", "job4")),
as.data.frame(Employability$observed) %>%
set_names(c("Employability1", "Employability2",
"Employability3", "Employability4")),
as.data.frame(job$observed) %>%
set_names(c("Career1", "Career2", "Career3", "Career4")))
#Creating a vector with the Cronbach's alpha for each subject
CronAlpha <- c(
alpha(df %>%
select(job1, job2, job3, job4))$total$std.alpha,
alpha(df %>%
select(Employability1, Employability2,
Employability3, Employability4))$total$std.alpha,
alpha(df %>%
select(Career1, Career2,
Career3, Career4))$total$std.alpha)
#Calculating the mean for each subject, than the correlations
Correlation <- df %>%
#Calculating the means
mutate(job = rowMeans(data.frame(job1, job2, job3, job4), na.rm = TRUE),
Employability =rowMeans(data.frame(Employability1, Employability2,
Employability3, Employability4), na.rm = TRUE),
Career =rowMeans(data.frame(Career1, Career2,
Career3, Career4), na.rm = TRUE)) %>%
#Selecting only the vars that I want for the correlation matrix
select(job, Employability, Career) %>%
as.matrix() %>%
rcorr()
#Extracting the Pearson's r
CorrelationRs <- Correlation$r
#Looping through the correlation data.frame and replacing with
# Cronbach's alpha
i <- 1
for (i in 1:nrow(CorrelationRs)) {
CorrelationRs[i, i] <- CronAlpha[i]
}
CorrelationRs
Edit
Instead of using loop, I should use diag().
diag(CorrelationRs) <- CronAlpha