Table 3.

Differences between the R psych and SPSS implementations

Procedure	Setting	R psych	SPSS	Note
PAF	Communality method	SMC, if this fails: unity	SMC, if this fails: MAC, if this fails, unity	How the diagonal of the original matrix is replaced to find initial eigenvalues
	Absolute eigenvalues	No	Yes	To avoid negative eigenvalues, SPSS takes the absolute of initial eigenvalues. This is not done in R psych, where negative eigenvalues might render the use of SMCs impossible
	Criterion type	Difference in sum of communalities	Difference in maximum individual communalities	Value on which the convergence criterion is applied
Promax	Varimax type	Singular value decomposition	Kaiser	SPSS follows the original varimax procedure from (Kaiser, 1958); likely with small changes in the varimax criterion), while R uses singular value decomposition
	Normalization of target matrix	Unnormalized	Normalized	The target matrix is row-normalized in SPSS, but not in R psych. This is not the Kaiser normalization, which is done in both implementations

A detailed description of the implementations of R psych and SPSS can be found in the supplemental material. PAF = principal axis factoring; SMC = squared multiple correlation; MAC = maximum absolute correlation, unity = all 1’s