That's just derived from an algebraic property...Remember, the model is that we model each observed variable (y) as a linear combination of unobserved factors (eta) and an error component (epsilon). So if we standardize (y), then the constituent components (eta, epsilon or factors, error) are also assumed to be standardized now.
thanks for your tutorials. They are very helpful for me!
Could you kindly explain why E[y]=0 implies that E[eta]=0 and E[epsilon]=0? Thanks in advance!
That's just derived from an algebraic property...Remember, the model is that we model each observed variable (y) as a linear combination of unobserved factors (eta) and an error component (epsilon). So if we standardize (y), then the constituent components (eta, epsilon or factors, error) are also assumed to be standardized now.
Maybe I didn't catch it, but are we also assuming that lambda is a constant so that we can pull it out of an expectation?