PROC FACTOR procedure

[Aline_A16]

Hello, 

i have a quick question : does the proc factor (method=prin) need prior standardization (mean=0 and std=1) of the continuous variables or by default is it included in the procedure? Actually, i am used to work with R software and for example the package FactoMineR with the PCA function performs a standardization by default.

Thank you in advance ! 

Aline

[PaigeMiller]

By default, PROC FACTOR works on correlation matrices. You can optionally request that it work on covariance matrices.

[Aline_A16]

I should go back to my statistics courses cause i forgot the difference between covariance and correlation matrix. Can you tell me in which case i need to perform a prior standardization ?

[PaigeMiller · Accepted solution]

Correlation comes from transforming variables to have mean=0 and variance=1. Covariance has mean=0, but no transformation of the variance is done.