I would like to obtain the initial communalities for an exploratory factor analysis in R (that is, the R squared of each item when predicted by the other items included in the analysis).
Is there a way to do this with either jmv::efa or psych::fa ?
I only see the uniqueness, which informs me of the communalities AFTER factor extraction (1-uniqueness)...
Thank you for your consideration : )
As you note, the initial communalities in a factor analysis are the squared multiple correlations (SMC) of each variable by the remaining variables. Using the built-in attitude dataset as an example they are easily calculated without additional packages via:
1 - 1 / diag(solve(cor(attitude)))
rating complaints privileges learning raises critical advance
0.7326020 0.7700868 0.3831176 0.6194561 0.6770498 0.1881465 0.5186447
The psych package includes the smc() function for convenience:
psych::smc(attitude)
rating complaints privileges learning raises critical advance
0.7326020 0.7700868 0.3831176 0.6194561 0.6770498 0.1881465 0.5186447