Reformat a correlation matrix in R to remove row names, column names, and cells in the top-right "half"

Question

To input a correlation matrix for SEM in R, I need to reformat a matrix in three ways:

• Remove row names
• Remove column names
• Remove all cells in the top-right "half" of the matrix, above the diagonal 1 values

In other words, the matrix...

          EDU       CLASS       INCOME    AUTONOMY        COL5         ALT5         IIDM
EDU       1.00000000 -0.14591135  0.246463592  0.21938208 -0.09727587 -0.043862753  0.062896738
CLASS    -0.14591135  1.00000000 -0.048511437 -0.01469029  0.01663138  0.036140650 -0.010155653
INCOME    0.24646359 -0.04851144  1.000000000  0.13778603 -0.06969773 -0.051922861  0.009402784
AUTONOMY  0.21938208 -0.01469029  0.137786034  1.00000000 -0.25223015  0.023174980  0.073941520
COL5     -0.09727587  0.01663138 -0.069697725 -0.25223015  1.00000000 -0.300825028 -0.062978212
ALT5     -0.04386275  0.03614065 -0.051922861  0.02317498 -0.30082503  1.000000000  0.001553936
IIDM      0.06289674 -0.01015565  0.009402784  0.07394152 -0.06297821  0.001553936  1.000000000

... needs to become the text...

 1.00000000
-0.14591135  1.00000000
 0.24646359 -0.04851144  1.000000000
 0.21938208 -0.01469029  0.137786034  1.00000000
-0.09727587  0.01663138 -0.069697725 -0.25223015  1.00000000
-0.04386275  0.03614065 -0.051922861  0.02317498 -0.30082503  1.000000000
 0.06289674 -0.01015565  0.009402784  0.07394152 -0.06297821  0.001553936  1.000000000

... to be inputted line-by-line into the readMoments function, as shown below.

R.bd <- readMoments(names=c('EDU',
                            'CLASS',
                            'INCOME',
                            'AUTONOMY',
                            'COL5',
                            'ALT5',
                            'IIDM'))
 1.00000000
-0.14591135  1.00000000
 0.24646359 -0.04851144  1.000000000
 0.21938208 -0.01469029  0.137786034  1.00000000
-0.09727587  0.01663138 -0.069697725 -0.25223015  1.00000000
-0.04386275  0.03614065 -0.051922861  0.02317498 -0.30082503  1.000000000
 0.06289674 -0.01015565  0.009402784  0.07394152 -0.06297821  0.001553936  1.000000000

We are currently doing this manually by copying, deleting, and pasting. Any suggestions on how to go about writing a function to do this would be greatly appreciated.

Thanks, Jonathan

Answer 1

Consider dat is your correlation matrix.

> dat <- as.matrix(dat)
> dimnames(dat) <- list(rep("", ncol(dat)), rep("", ncol(dat)))
> dat[upper.tri(dat)] <- NA
> print(dat, na.print = " ") 
                       # you can also make a replacement `dat[is.na(dat)] <- " "`

  1.00000000                                                               
 -0.14591135  1.00000000                                                   
  0.24646359 -0.04851144  1.000000000                                      
  0.21938208 -0.01469029  0.137786034  1.00000000                          
 -0.09727587  0.01663138 -0.069697725 -0.25223015  1.00000000              
 -0.04386275  0.03614065 -0.051922861  0.02317498 -0.30082503 1.000000000  
  0.06289674 -0.01015565  0.009402784  0.07394152 -0.06297821 0.001553936 1

Answer 2

You can create the object R.bd directly. You don't need to input line by line.

Assuming, dat is the name of the object (data frame or matrix):

dat[upper.tri(dat)] <- 0 # replace all elements above the diagonal with 0
R.bd <- as.matrix(dat)  # transform to matrix (not necessary if it's a matrix)

That's it. You don't need to specify the names since they are already in the matrix object.